π Key Insights β 40-Month Analysis Takeaways
Headline findings from the full dashboard analysis, grouped into a narrative arc. Each card points to the tab where the underlying data lives β every claim here is auditable against the raw data.
π» The Headline Problem β Unit Economics Are Eroding
2 findings
NetSuite ROAS has collapsed 9.84x β 5.09x
Every paid dollar produced nearly 2Γ as much NetSuite revenue three years ago as it does today (2023 β Q1 '26). The single most condensed metric showing unit economics have deteriorated.
β³ Overview Β· Multi-Year Context
Every paid dollar produced nearly 2Γ as much NetSuite revenue three years ago as it does today (2023 β Q1 '26). The single most condensed metric showing unit economics have deteriorated.
β³ Overview Β· Multi-Year Context
Spend +28% YoY, revenue β21% YoY (Q1 '26)
Paying more, getting less β both in absolute and relative terms. The efficiency problem is sharper than either metric alone shows.
β³ Overview Β· Quad Anchor + YoY table
Paying more, getting less β both in absolute and relative terms. The efficiency problem is sharper than either metric alone shows.
β³ Overview Β· Quad Anchor + YoY table
β
Campaign Efficiency Is Strong β Allocation Is the Issue
4 findings Β· the most important section
Brand ROAS improved every year: 3.80x β 5.54x β 8.26x
Brand is NOT the efficiency problem. The concern is allocation and cannibalization, not campaign-level execution.
β³ Brand vs NB Β· 4-year trajectory
Brand is NOT the efficiency problem. The concern is allocation and cannibalization, not campaign-level execution.
β³ Brand vs NB Β· 4-year trajectory
Non-Brand ROAS also compounded: 0.94x β 1.53x β 2.54x
the Client's own data shows NB scales get more efficient with sustained investment (2.7Γ improvement over 3 years). NB still runs below Brand efficiency β that's expected; NB creates demand, Brand harvests it.
β³ Brand vs NB Β· 4-year trajectory
the Client's own data shows NB scales get more efficient with sustained investment (2.7Γ improvement over 3 years). NB still runs below Brand efficiency β that's expected; NB creates demand, Brand harvests it.
β³ Brand vs NB Β· 4-year trajectory
Brand-heavy months β NB ROAS drops 3 months later (r = β0.63)
Strongest lag signal in the entire 40-month dataset. Predicts continued NB/Brand compression through MayβJul '26 if current allocation continues.
β³ Lag Effects Β· Brand share β NB ROAS T+3
Strongest lag signal in the entire 40-month dataset. Predicts continued NB/Brand compression through MayβJul '26 if current allocation continues.
β³ Lag Effects Β· Brand share β NB ROAS T+3
Q1 '26 Search IS at 85% is a measurement artifact, not a win
96β100% of that figure comes from Brand campaigns. NB Search IS has never crossed 14% in any campaign-month across 40 months. It's a symptom of the NB pullback + match-type transition.
β³ Brand vs NB Β· Search IS as Supporting Signal
96β100% of that figure comes from Brand campaigns. NB Search IS has never crossed 14% in any campaign-month across 40 months. It's a symptom of the NB pullback + match-type transition.
β³ Brand vs NB Β· Search IS as Supporting Signal
π Statistical Patterns From the 40-Month Dataset
2 findings Β· how the data actually behaves
+0.69 is the strongest correlation in the dataset
Same-month Total Paid Spend β NetSuite Revenue. All lagged spend correlations drop to ~0. The Client is direct-response-dominant β cuts hurt this month, not future months.
β³ Lag Effects Β· Pearson table
Same-month Total Paid Spend β NetSuite Revenue. All lagged spend correlations drop to ~0. The Client is direct-response-dominant β cuts hurt this month, not future months.
β³ Lag Effects Β· Pearson table
Promo pull-forward is statistically real (r = β0.32 at T+2)
Heavy-discount months are followed by softer months 2 months later. 4-year pattern β worth building into the promo calendar for Q4 '26.
β³ Product Sales Β· Discount correlation
Heavy-discount months are followed by softer months 2 months later. 4-year pattern β worth building into the promo calendar for Q4 '26.
β³ Product Sales Β· Discount correlation
β οΈ Operational & Structural Changes
2 findings Β· recent shifts in how the business operates
Attach rate exactly 1.0 on every order since Feb 1, 2026
Down from 3-year baseline of ~1.5. Single-SKU orders only β the most precise product-mix discontinuity in the Client's history. Operational, not demand-driven.
β³ Product Sales Β· Attach rate & bundling
Down from 3-year baseline of ~1.5. Single-SKU orders only β the most precise product-mix discontinuity in the Client's history. Operational, not demand-driven.
β³ Product Sales Β· Attach rate & bundling
PMax spend nearly tripled 2023 β 2025 ($83K β $280K), now reversing
PMax was the 2025 growth engine. The current PMax pullback is mechanically behind most of the "NB spend cut" narrative β one decision, not two.
β³ More Reports β Campaign Type Β· 4-year evolution
PMax was the 2025 growth engine. The current PMax pullback is mechanically behind most of the "NB spend cut" narrative β one decision, not two.
β³ More Reports β Campaign Type Β· 4-year evolution
π‘ Untapped Opportunities
2 findings Β· levers the data says could move revenue
Apr '26 Meta Earth Day RT hit 3.49x β highest RMKT ROAS ever
40-month structural gap (RMKT 1.9% of spend vs 15β30% DTC benchmark) may now have a working model to scale against.
β³ More Reports β RMKT vs Prosp
40-month structural gap (RMKT 1.9% of spend vs 15β30% DTC benchmark) may now have a working model to scale against.
β³ More Reports β RMKT vs Prosp
Organic Search has not compounded despite 3+ years of paid investment
Flat at 7β10% of sessions across the entire GA4 window. In healthy DTC mattress accounts, Organic should grow to 15β25% as brand awareness builds. Consistent with the cannibalization read.
β³ More Reports β Traffic Source
Flat at 7β10% of sessions across the entire GA4 window. In healthy DTC mattress accounts, Organic should grow to 15β25% as brand awareness builds. Consistent with the cannibalization read.
β³ More Reports β Traffic Source
How to use this view: These 12 findings are the core takeaways of a 40-month analysis spanning Google / Meta / Bing paid media, NetSuite sales, and GA4 traffic. Each points to the tab where the underlying data, charts, and supporting correlations live β every claim is auditable.
For the context behind them, start with the Overview tab (40-month NetSuite revenue + NS-ROAS trend), then the Lag Effects tab for the correlation table, then Brand vs NB and Product Sales for the structural detail. The Recommendations tab translates these findings into a course-correction plan.
For the context behind them, start with the Overview tab (40-month NetSuite revenue + NS-ROAS trend), then the Lag Effects tab for the correlation table, then Brand vs NB and Product Sales for the structural detail. The Recommendations tab translates these findings into a course-correction plan.
π― Recommendations β Course Correction Plan
Recommendations grounded in the 40-month dataset. Each action ties to specific findings on the Key Insights tab and the supporting drivers throughout the dashboard.
π¨ Immediate (next 30 days) β stop the bleeding
2 actions Β· directly addresses the Q1 '26 efficiency erosion
1. Rebalance to a 70% Non-Brand / 30% Brand spend split β across the full NB toolkit, not just one channel
Where we are: Q1 '26 paid-search spend split is roughly 48% NB / 52% Brand (bidirectionally adjusted) β reversed from a healthy 2025 ratio. Reference the 4-year arc:
Target: Move back toward 70% Non-Brand / 30% Brand. 2025 hit 64/36 (the highest NB share the Client has ever run) with NB ROAS at 2.54x and Brand at 8.26x β the lift comes from doing more of what was already working. Continuing to push to 70/30 is a directional stretch beyond 2025's best.
How NB spend should be deployed (not just PMax): the NB pool is bigger than one channel. Spread the increased investment across the full prospecting toolkit:
Β· Meta (Sales + upper-funnel) β 40-month T+0 correlation with NS revenue is +0.49
Β· PMax (NonBrand variants) β restore catalog-driven NB targeting
Β· Shopping β recently launched (Mar '26), early ROAS data is encouraging
Β· YouTube / Demand Gen β upper-funnel awareness layer
Β· Search NonBrand β pure intent capture
Continuous improvement on classification: keep adding negative keywords in BOTH directions β to Brand campaigns to suppress non-brand-name queries (~21% leak rate, structural across all 4 years), AND to NB campaigns to keep brand-name queries out (~6% leak rate Q1 '26, already improving year-over-year). This protects the integrity of the 70/30 measurement.
Β· Same-month Total Paid Spend β NS Revenue: r = +0.68 (direct-response dominant)
Β· Brand-heavy months β NB ROAS T+3: r = β0.63 (current allocation predicts further compression)
Β· Brand-heavy months β Brand ROAS T+3: r = β0.61 (Brand also compresses, not just NB)
Β· NB ROAS improved every year with sustained investment: 0.94x β 1.53x β 2.54x
Β· Brand ROAS improved as NB share grew: 3.80x β 5.54x β 8.26x β evidence that letting NB do the demand-creation work makes Brand more efficient, not less
Where we are: Q1 '26 paid-search spend split is roughly 48% NB / 52% Brand (bidirectionally adjusted) β reversed from a healthy 2025 ratio. Reference the 4-year arc:
| Year | NB share | Brand share | Read |
|---|---|---|---|
| 2023 | 44% | 56% | Brand-heavy starting point |
| 2024 | 51% | 49% | NB scaling, healthy mix shift |
| 2025 | 64% | 36% | Healthiest mix the Client has run Β· best NB ROAS year |
| Q1 '26 | 48% | 52% | Reversed β Brand back to a 2023-like share |
How NB spend should be deployed (not just PMax): the NB pool is bigger than one channel. Spread the increased investment across the full prospecting toolkit:
Β· Meta (Sales + upper-funnel) β 40-month T+0 correlation with NS revenue is +0.49
Β· PMax (NonBrand variants) β restore catalog-driven NB targeting
Β· Shopping β recently launched (Mar '26), early ROAS data is encouraging
Β· YouTube / Demand Gen β upper-funnel awareness layer
Β· Search NonBrand β pure intent capture
Continuous improvement on classification: keep adding negative keywords in BOTH directions β to Brand campaigns to suppress non-brand-name queries (~21% leak rate, structural across all 4 years), AND to NB campaigns to keep brand-name queries out (~6% leak rate Q1 '26, already improving year-over-year). This protects the integrity of the 70/30 measurement.
π§ The mental model: more attributed revenue with LESS Brand spend
The intuition that "more Brand spend = more brand revenue" is backwards in a mature account. Three structural reasons less Brand spend produces a healthier business:
(1) Brand efficiency depreciates as Brand spend rises. When you bid harder on a finite pool of branded queries, you're paying more per click for traffic that was already brand-loyal. Brand campaigns harvest existing intent β they don't create new demand. Heavier Brand investment compresses Brand ROAS over time (the Lag Effects T+3 β0.61 signal). Lighter Brand spend captures the same converters at a lower cost.
(2) Heavy Brand paid spend cannibalizes non-paid revenue. Most of the people typing "[brand name]" into Google were already going to land on the site β through organic search, direct, email, or referral. When the Client bids aggressively on those queries, the conversion still happens but moves from "free" channels to "paid Brand." NS revenue stays the same; ROAS calculations look better; but the actual incremental revenue from the Brand spend is small. The Brand traffic share rising 7-10pp without total non-CPC traffic rising in Q1 '26 is the textbook fingerprint of this.
(3) Lower Brand share = more investment in brand DEVELOPMENT, not less. This is the counterintuitive part: NB spend (Search NB, Shopping, Meta, YouTube, Demand Gen) is what introduces new people to the brand. Sustained NB investment grows the awareness pool β which then naturally feeds Brand campaigns more efficient converters in subsequent months. NB IS brand investment. Holding NB at 70% of paid keeps the awareness funnel filling, which compounds: more new shoppers β bigger brand-aware audience β higher Brand campaign ROAS at the same Brand budget. The Client's own 3-year arc proves it (Brand ROAS 3.80x β 8.26x as NB scaled).
The takeaway: the 70/30 NB/Brand target is not "spend less on the brand." It's "spend on the levers that build the brand long-term, and let Brand campaigns do their job β closing demand at scale β without inflating their budget past where the demand actually lives."
Why this works (data evidence):The intuition that "more Brand spend = more brand revenue" is backwards in a mature account. Three structural reasons less Brand spend produces a healthier business:
(1) Brand efficiency depreciates as Brand spend rises. When you bid harder on a finite pool of branded queries, you're paying more per click for traffic that was already brand-loyal. Brand campaigns harvest existing intent β they don't create new demand. Heavier Brand investment compresses Brand ROAS over time (the Lag Effects T+3 β0.61 signal). Lighter Brand spend captures the same converters at a lower cost.
(2) Heavy Brand paid spend cannibalizes non-paid revenue. Most of the people typing "[brand name]" into Google were already going to land on the site β through organic search, direct, email, or referral. When the Client bids aggressively on those queries, the conversion still happens but moves from "free" channels to "paid Brand." NS revenue stays the same; ROAS calculations look better; but the actual incremental revenue from the Brand spend is small. The Brand traffic share rising 7-10pp without total non-CPC traffic rising in Q1 '26 is the textbook fingerprint of this.
(3) Lower Brand share = more investment in brand DEVELOPMENT, not less. This is the counterintuitive part: NB spend (Search NB, Shopping, Meta, YouTube, Demand Gen) is what introduces new people to the brand. Sustained NB investment grows the awareness pool β which then naturally feeds Brand campaigns more efficient converters in subsequent months. NB IS brand investment. Holding NB at 70% of paid keeps the awareness funnel filling, which compounds: more new shoppers β bigger brand-aware audience β higher Brand campaign ROAS at the same Brand budget. The Client's own 3-year arc proves it (Brand ROAS 3.80x β 8.26x as NB scaled).
The takeaway: the 70/30 NB/Brand target is not "spend less on the brand." It's "spend on the levers that build the brand long-term, and let Brand campaigns do their job β closing demand at scale β without inflating their budget past where the demand actually lives."
Β· Same-month Total Paid Spend β NS Revenue: r = +0.68 (direct-response dominant)
Β· Brand-heavy months β NB ROAS T+3: r = β0.63 (current allocation predicts further compression)
Β· Brand-heavy months β Brand ROAS T+3: r = β0.61 (Brand also compresses, not just NB)
Β· NB ROAS improved every year with sustained investment: 0.94x β 1.53x β 2.54x
Β· Brand ROAS improved as NB share grew: 3.80x β 5.54x β 8.26x β evidence that letting NB do the demand-creation work makes Brand more efficient, not less
2. Reactivate cart-value upsell mechanics β and clean up the underlying data with the the client team
What we observed: Attach rate dropped from a 3-year baseline of ~1.5 to exactly 1.0 on every order since Feb 1, 2026. AOV fell from $747 (Jan '25) to $486 (Mar '26).
Two-track action:
(a) Data validation with the Client β open a working session with their team to confirm whether this is a real bundling change, an e-commerce site issue (broken cross-sell flow), or a NetSuite SKU labeling change. The "exactly 1.0" precision on every order suggests it could be measurement, not behavior.
(b) Push for cart-value uplift regardless β even if the data resolution comes back inconclusive, the directional signal is strong: months with higher add-on values per order have consistently been the Client's best-performing months. Reactivate cross-sell on PDPs, post-add-to-cart upsell modules, mattress + accessory bundling at checkout.
Why this matters: AOV is currently ~$170 (~25%) below the 3-year baseline. If half of that is recoverable through upsell mechanics, that's ~$85/order Γ ~500 orders/mo = ~$42K/month of recoverable AOV β independent of any media-side action.
What we observed: Attach rate dropped from a 3-year baseline of ~1.5 to exactly 1.0 on every order since Feb 1, 2026. AOV fell from $747 (Jan '25) to $486 (Mar '26).
Two-track action:
(a) Data validation with the Client β open a working session with their team to confirm whether this is a real bundling change, an e-commerce site issue (broken cross-sell flow), or a NetSuite SKU labeling change. The "exactly 1.0" precision on every order suggests it could be measurement, not behavior.
(b) Push for cart-value uplift regardless β even if the data resolution comes back inconclusive, the directional signal is strong: months with higher add-on values per order have consistently been the Client's best-performing months. Reactivate cross-sell on PDPs, post-add-to-cart upsell modules, mattress + accessory bundling at checkout.
Why this matters: AOV is currently ~$170 (~25%) below the 3-year baseline. If half of that is recoverable through upsell mechanics, that's ~$85/order Γ ~500 orders/mo = ~$42K/month of recoverable AOV β independent of any media-side action.
π Medium-term (next 60β90 days) β fix the structure
2 actions Β· build the longer-cycle levers
3. Scale Meta Remarketing β segmented by audience proximity
Anchor evidence: April '26 Meta Earth Day RT campaign hit 3.49x ROAS β the highest RMKT ROAS in the entire 40-month record. RMKT has been chronically under-invested at 1.9% of spend (2025 avg) vs the 15β30% DTC benchmark.
Audience segmentation (the messaging lift): rather than one generic RT campaign, structure tiers based on user proximity:
Β· Never visited / never purchased β full brand-introduction creative; lead with the organic mattress story and certifications
Β· Visited but not purchased β push them over the line: comfort guarantee, financing options, current promo
Β· Long-dormant past purchasers β repurchase / accessory cross-sell Β· upgrade messaging Β· referrals
Honest read on data evidence: Pearson correlation of RMKT spend β NS revenue is weak (T+0 +0.31) β but that's because the Client has never run RMKT at scale. At 1.9% of spend, the data can't isolate RMKT's direct lift; the correlation is mostly RMKT co-moving with prospecting in promo windows. The Apr '26 Meta RT spike is the first concrete proof that audience + creative + timing actually works at this account. Scaling it IS the test β there's no way to know if RMKT lifts NS revenue without first running it at the level the structural argument suggests it should be at.
Target: grow RMKT share of paid from 1.9% β 8β15% over the next 90 days β still below the DTC benchmark but a credible first step. Mirror the Meta RT structure into Google Display RT to replace the legacy Display-Remarketing campaign that has run at 0.21x ROAS for years.
Anchor evidence: April '26 Meta Earth Day RT campaign hit 3.49x ROAS β the highest RMKT ROAS in the entire 40-month record. RMKT has been chronically under-invested at 1.9% of spend (2025 avg) vs the 15β30% DTC benchmark.
Audience segmentation (the messaging lift): rather than one generic RT campaign, structure tiers based on user proximity:
Β· Never visited / never purchased β full brand-introduction creative; lead with the organic mattress story and certifications
Β· Visited but not purchased β push them over the line: comfort guarantee, financing options, current promo
Β· Long-dormant past purchasers β repurchase / accessory cross-sell Β· upgrade messaging Β· referrals
Honest read on data evidence: Pearson correlation of RMKT spend β NS revenue is weak (T+0 +0.31) β but that's because the Client has never run RMKT at scale. At 1.9% of spend, the data can't isolate RMKT's direct lift; the correlation is mostly RMKT co-moving with prospecting in promo windows. The Apr '26 Meta RT spike is the first concrete proof that audience + creative + timing actually works at this account. Scaling it IS the test β there's no way to know if RMKT lifts NS revenue without first running it at the level the structural argument suggests it should be at.
Target: grow RMKT share of paid from 1.9% β 8β15% over the next 90 days β still below the DTC benchmark but a credible first step. Mirror the Meta RT structure into Google Display RT to replace the legacy Display-Remarketing campaign that has run at 0.21x ROAS for years.
4. Build Organic Search as a parallel investment lever
The gap: Organic Search has been flat at 7β10% of GA4 sessions across the entire measurement window (Feb '25 β Apr '26). In healthy DTC mattress accounts, Organic typically grows to 15β25% as brand awareness compounds. The Client has 3+ years of paid brand investment behind it β Organic should be compounding. It isn't, which is consistent with the cannibalization read on the Brand vs NB tab (paid brand is capturing what could be organic growth).
Why this matters strategically: Organic Search is compounding β every new Organic visitor reduces the brand-search-paid bill in subsequent months. It's the one channel that gets cheaper over time when invested in correctly.
Action plan:
Β· SEO content investment β long-tail organic mattress queries (where Avocado / Birch / Saatva / Naturepedic dominate): "best organic mattress for [persona]," "organic latex vs memory foam," "GOLS-certified mattresses," etc.
Β· Technical SEO audit β check indexing, Core Web Vitals, schema, internal linking
Β· Topical authority build β mattress-buying education content, organic-certification deep-dives
Β· Pair with the NB rebalance (action #1) β as Brand spend share comes down, give Organic the room to claim some of that branded-intent traffic naturally
Note: this is a long-cycle play (6+ months to compound) but compounds permanently. It belongs in the recommendation set even though it pays back slower than #1β3.
The gap: Organic Search has been flat at 7β10% of GA4 sessions across the entire measurement window (Feb '25 β Apr '26). In healthy DTC mattress accounts, Organic typically grows to 15β25% as brand awareness compounds. The Client has 3+ years of paid brand investment behind it β Organic should be compounding. It isn't, which is consistent with the cannibalization read on the Brand vs NB tab (paid brand is capturing what could be organic growth).
Why this matters strategically: Organic Search is compounding β every new Organic visitor reduces the brand-search-paid bill in subsequent months. It's the one channel that gets cheaper over time when invested in correctly.
Action plan:
Β· SEO content investment β long-tail organic mattress queries (where Avocado / Birch / Saatva / Naturepedic dominate): "best organic mattress for [persona]," "organic latex vs memory foam," "GOLS-certified mattresses," etc.
Β· Technical SEO audit β check indexing, Core Web Vitals, schema, internal linking
Β· Topical authority build β mattress-buying education content, organic-certification deep-dives
Β· Pair with the NB rebalance (action #1) β as Brand spend share comes down, give Organic the room to claim some of that branded-intent traffic naturally
Note: this is a long-cycle play (6+ months to compound) but compounds permanently. It belongs in the recommendation set even though it pays back slower than #1β3.
π‘ Leading Indicators β Monthly Review Dashboard
7 metrics to track monthly Β· each tied to a course-correction recommendation
Watch these 7 metrics on every monthly review. Each has a historical pattern in the 40-month dataset that determines what "healthy" looks like and when to escalate.
| Metric | Healthy | Warning | Action trigger | Source |
|---|---|---|---|---|
| Brand share of paid-search spend | 25β40% | 40β50% | >50% for 2 consecutive months | Brand vs NB tab |
| Non-Brand ROAS (platform-reported) | 2.3x+ | 2.0β2.3x | <2.0x = NB scaling without efficiency floor Β· pause and audit creative/audiences | Brand vs NB Β· Segment ROAS |
| Brand ROAS (platform-reported) | 6.5x+ | 5.0β6.5x | <5.0x = T+3 lag effect feeding through (NB cuts catching up) | Brand vs NB Β· Segment ROAS |
| Total GA4 sessions per month (3-mo trailing avg) | β₯ 46K/mo Β· trending up | 40β46K/mo Β· flat | < 38K/mo = NB investment not generating volume | Traffic Source tab Β· all channels |
| Traffic quality β engagement time, bounce rate, views/session (grouped) | Eng. time: 26β29s Bounce: 52β58% Views/session: 2.19β2.42 |
>10% deviation on any metric | Sustained >10% deviation 2+ consecutive months, or multiple metrics drifting simultaneously | Traffic Source tab |
| Attach rate (line items per order) | 1.4+ | 1.2β1.4 | <1.2 = upsell mechanics broken (current state) | Product Sales Β· Attach |
| Total NetSuite revenue YoY | Flat or positive | β5% to β15% | <β15% sustained 2+ months = full strategic review | Overview Β· YoY chart |
How to use this table: these aren't aspirational targets β they're tripwires built from the 40-month historical pattern.
On the NB ROAS floor (2.3x): as NB scales toward the 70/30 target, expect ROAS to compress somewhat β that's normal because adding spend means reaching less-converted audiences. The 2.3x floor is set just below 2025's 2.54x annual NB ROAS to allow room for scaling without falling out of efficient territory. If NB ROAS drops below 2.0x while NB spend is rising, the scaling isn't working β pause and audit (creative, audience targeting, exclusions) before continuing to scale.
On the Brand share guardrail (25β40%): intentionally tighter than the previous 35β50% draft because the 40-month evidence shows NB ROAS compression starts as soon as Brand share crosses ~50% and lasts 2β3 months. Catching this 1 month earlier is the difference between protective adjustment and a quarter-long correction.
On traffic quality (the grouped row): volume growth is meaningless if it's bringing in shittier traffic. Engagement time, bounce rate, and views/session are interdependent β they move as a group. Track all three against the current baseline. If sessions grow but quality drops, NB scaling is buying volume at the cost of intent (audience targeting issue, creative-landing mismatch, or budget pushed into low-quality placements). Reference baselines from the recent 3-month trailing average:
Healthy: each metric within Β±5% of these baselines (e.g. paid eng. time 26.3β29.1s, paid bounce 52.4β57.9%). Track paid and non-paid separately β paid quality is a creative/targeting signal; non-paid quality is a brand/site signal.
On the NB ROAS floor (2.3x): as NB scales toward the 70/30 target, expect ROAS to compress somewhat β that's normal because adding spend means reaching less-converted audiences. The 2.3x floor is set just below 2025's 2.54x annual NB ROAS to allow room for scaling without falling out of efficient territory. If NB ROAS drops below 2.0x while NB spend is rising, the scaling isn't working β pause and audit (creative, audience targeting, exclusions) before continuing to scale.
On the Brand share guardrail (25β40%): intentionally tighter than the previous 35β50% draft because the 40-month evidence shows NB ROAS compression starts as soon as Brand share crosses ~50% and lasts 2β3 months. Catching this 1 month earlier is the difference between protective adjustment and a quarter-long correction.
On traffic quality (the grouped row): volume growth is meaningless if it's bringing in shittier traffic. Engagement time, bounce rate, and views/session are interdependent β they move as a group. Track all three against the current baseline. If sessions grow but quality drops, NB scaling is buying volume at the cost of intent (audience targeting issue, creative-landing mismatch, or budget pushed into low-quality placements). Reference baselines from the recent 3-month trailing average:
| Segment | Engagement time | Bounce rate | Views/session |
|---|---|---|---|
| Paid (CPC) | 27.7s | 55.1% | 2.30 |
| Non-Paid (Direct/Organic/Email/etc.) | 31.7s | 54.4% | 2.30 |
π GA4 traffic growth projection β what NB investment should produce
The NB rebalance is supposed to fund top-of-funnel demand creation. The way we'll know it's working is if total GA4 traffic β paid AND non-paid combined β starts growing at a moderate, sustained pace.
Where total GA4 sessions are today (sessions per month):
Projected growth path as NB rebalance compounds (aggressive β Apr '26 is the new starting point, not the ceiling). All figures are 3-month trailing average of monthly sessions:
Why the targets are aggressive: April '26 (full month, first month with the Brand-vs-NB rebalance starting) hit 41,959 sessions with 33,484 users across paid + non-paid β exceeding 2025 H2 levels and tying the highest non-promo session month in the 13-month GA4 window. With the NB rebalance compounding through Q2 and into the summer mattress-demand window, July should be +8β10% above Apr '26 at minimum. By BFCM (Oct), we should be 25%+ above Apr's level as paid and organic both compound. By Apr '27 β a full year of sustained NB investment β we should be running above 2025's peak month (Aug at 48.6K) on a sustained basis, setting a new traffic ceiling for the brand. Apr '26 is the starting line, not the destination.
Why monthly (not quarterly): per-month visibility surfaces seasonal variation (BFCM, post-holiday lull, summer mattress season) that quarterly totals would smooth over. The 3-mo trailing average filters noise from single-month spikes (Memorial Day, July 4, Labor Day). Track the trailing avg every month; review the trajectory each quarter. Why the trajectory matters more than the absolute number: if NB spend goes up but sessions stay flat, the NB investment is reaching the same audience pool β that's a creative/targeting issue. If sessions grow but only paid sessions grow (organic/direct stay flat), the spend is buying volume but not building the brand. The healthy signal is sessions growing across BOTH paid AND non-paid simultaneously β that's NB doing its actual job (introducing new users who then return via direct/organic later).
What to track alongside total sessions: total users (cohort growth), share of returning vs new users (NB drives new-user share up), and Organic Search session growth specifically (the cleanest signal of compounding brand awareness β Recommendation #4).
Where total GA4 sessions are today (sessions per month):
| Window | Avg sessions/month | Read |
|---|---|---|
| 2025 H2 (JulβDec) | ~36K/mo | Reference healthy state Β· NB at full scale Β· 49K Aug peak |
| Q4 '25 (OctβDec) | ~33K/mo | BFCM peak (Dec 43K) Β· normal seasonal pattern |
| JanβMar '26 avg | ~30K/mo | Post-NB-cut trough Β· sessions softened |
| Apr '26 (full month) | ~42K/mo | Recovery β Apr exceeded 2025 H2 baseline as Brand vs NB rebalance began Β· highest non-promo month in 13-mo window |
| 3-mo trailing avg (FebβApr '26) | ~35K/mo | Already past the 35K healthy threshold Β· trending up sharply |
| Horizon | Target sessions/month | vs Apr '26 (42K) | What this implies |
|---|---|---|---|
| 90 days (Jul '26) | β₯ 46K/mo | +10% | NB rebalance + summer mattress demand Β· Q3 should outperform Q1, not just match Apr |
| 6 months (Oct '26) | β₯ 52K/mo | +25% | Pre-BFCM ramp Β· NB compounding into the high-intent window Β· paid AND organic growing together |
| 12 months (Apr '27) | β₯ 63K/mo | +50% | Sustained above 2025's all-time peak (49K Aug) Β· new ceiling for the brand |
Why monthly (not quarterly): per-month visibility surfaces seasonal variation (BFCM, post-holiday lull, summer mattress season) that quarterly totals would smooth over. The 3-mo trailing average filters noise from single-month spikes (Memorial Day, July 4, Labor Day). Track the trailing avg every month; review the trajectory each quarter. Why the trajectory matters more than the absolute number: if NB spend goes up but sessions stay flat, the NB investment is reaching the same audience pool β that's a creative/targeting issue. If sessions grow but only paid sessions grow (organic/direct stay flat), the spend is buying volume but not building the brand. The healthy signal is sessions growing across BOTH paid AND non-paid simultaneously β that's NB doing its actual job (introducing new users who then return via direct/organic later).
What to track alongside total sessions: total users (cohort growth), share of returning vs new users (NB drives new-user share up), and Organic Search session growth specifically (the cleanest signal of compounding brand awareness β Recommendation #4).
π Meta Scale-Up Benchmarks β $13K β $20K/month
Platform-diversification play Β· 6 metrics to monitor as Meta spend ramps
Why scale Meta now: Meta efficiency has structurally improved across 2025-26 β cost per session dropped from $5.30 (Aug '25) to $1.62 (Apr '26), a 69% improvement. Meta-driven sessions grew from 1.5K (Aug '25) to ~8K (Apr '26) as spend scaled, and non-CPC traffic has been creeping up alongside Meta investment β a halo effect suggesting Meta is helping introduce the brand to new audiences who later return via direct / organic. The platform is also a diversification play (Google ran 78% of paid spend in Q1 '26; over-concentration risk).
Target trajectory: Apr '26 spend was $12.9K. Inch toward $20K/month over the next 60β90 days (~+55%). At Apr '26 efficiency, that's roughly +11.5K Meta-driven users / +12.3K sessions per month on top of current volumes.
Target trajectory: Apr '26 spend was $12.9K. Inch toward $20K/month over the next 60β90 days (~+55%). At Apr '26 efficiency, that's roughly +11.5K Meta-driven users / +12.3K sessions per month on top of current volumes.
| Metric | Apr '26 baseline | Healthy as we scale | Warning | Action trigger |
|---|---|---|---|---|
| Meta spend / month | $12.9K | $18β20K Β· stepping up | $13β18K Β· ramp slower than expected | <$13K = stalled / decelerating |
| Cost per session (CPS) | $1.62 | β€ $2.00 | $2.00β2.50 | > $2.50 = scaling into less-efficient inventory Β· pause and audit |
| Cost per user (unique) | $1.74 | β€ $2.50 | $2.50β3.00 | > $3.00 = audience saturation signal |
| Meta-driven users / month | 7,446 | β₯ 10K Β· trending up with spend | 8β10K Β· spend up but volume flat | < 8K despite higher spend = creative fatigue or audience exhaustion |
| Engaged session rate | 25.2% | β₯ 25% | 22β25% | < 22% = traffic quality dropping as we scale (creative or targeting) |
| Total non-CPC sessions (halo signal) | ~20K | Trending up alongside Meta scale | Flat while Meta spend grows | Declining = Meta growth not building the broader funnel |
How these benchmarks fit together:
(1) Spend & volume should move together. If Meta spend hits $20K but Meta-driven users stay flat at 7-8K, the additional spend is buying frequency on the same audience (saturation), not reach. CPS will rise as you push past audience capacity β that's the leading indicator. If CPS holds <$2.00 while spend scales, audiences still have headroom.
(2) Engaged session rate is the quality floor. Apr '26 ran 25% engaged β keep it there. If engagement drops below 22% as Meta scales, the campaigns are reaching less-qualified audiences (broader targeting, weaker creative match, or wrong placements). Pause scaling, audit creative + audience, then resume.
(3) Non-CPC halo is the strategic signal. Meta's real value isn't just direct conversions (Meta in-platform ROAS is structurally under-attributed since iOS 14). The signal that Meta is working is whether total non-CPC traffic is rising as Meta spend rises. If Meta scales to $20K and non-CPC stays flat, Meta is buying paid sessions but not building brand awareness. If non-CPC grows alongside Meta, Meta is doing the demand-creation work the recommendation predicts.
Specific Meta share-of-paid target: as Meta climbs to $20K and Google + Bing hold roughly steady, Meta share of total paid moves from ~31% (Apr '26) toward 40β45% within 90 days. That's a healthier diversification β still Google-dominant, but with Meta as a real second pillar rather than a supplemental channel.
(1) Spend & volume should move together. If Meta spend hits $20K but Meta-driven users stay flat at 7-8K, the additional spend is buying frequency on the same audience (saturation), not reach. CPS will rise as you push past audience capacity β that's the leading indicator. If CPS holds <$2.00 while spend scales, audiences still have headroom.
(2) Engaged session rate is the quality floor. Apr '26 ran 25% engaged β keep it there. If engagement drops below 22% as Meta scales, the campaigns are reaching less-qualified audiences (broader targeting, weaker creative match, or wrong placements). Pause scaling, audit creative + audience, then resume.
(3) Non-CPC halo is the strategic signal. Meta's real value isn't just direct conversions (Meta in-platform ROAS is structurally under-attributed since iOS 14). The signal that Meta is working is whether total non-CPC traffic is rising as Meta spend rises. If Meta scales to $20K and non-CPC stays flat, Meta is buying paid sessions but not building brand awareness. If non-CPC grows alongside Meta, Meta is doing the demand-creation work the recommendation predicts.
Specific Meta share-of-paid target: as Meta climbs to $20K and Google + Bing hold roughly steady, Meta share of total paid moves from ~31% (Apr '26) toward 40β45% within 90 days. That's a healthier diversification β still Google-dominant, but with Meta as a real second pillar rather than a supplemental channel.
β±οΈ Sequencing β why this order
The order matters because of the lag relationships in the data:
Β· Months 1β2: rebalance to 70/30 NB/Brand (#1) and reactivate upsell (#2). Both have direct same-month NS revenue effect (the +0.69 correlation).
Β· Months 2β4: Meta RT scale-up (#3) shows efficacy in 4β6 weeks once the audience tiers are built and creative is in market.
Β· Months 3β6+: Organic SEO (#4) starts compounding. The Brand-share lag effect from current allocation also surfaces here β restoring NB now front-loads protection against the predicted MayβJul '26 compression.
Doing #1 alone protects current-month revenue. Doing #1 + #3 protects current AND future months. Adding #4 builds compounding leverage that reduces dependency on paid brand long-term.
Β· Months 1β2: rebalance to 70/30 NB/Brand (#1) and reactivate upsell (#2). Both have direct same-month NS revenue effect (the +0.69 correlation).
Β· Months 2β4: Meta RT scale-up (#3) shows efficacy in 4β6 weeks once the audience tiers are built and creative is in market.
Β· Months 3β6+: Organic SEO (#4) starts compounding. The Brand-share lag effect from current allocation also surfaces here β restoring NB now front-loads protection against the predicted MayβJul '26 compression.
Doing #1 alone protects current-month revenue. Doing #1 + #3 protects current AND future months. Adding #4 builds compounding leverage that reduces dependency on paid brand long-term.
β Open Questions for the Client's Team β Context We Need to Sharpen the Recommendations
The dashboard analysis runs against paid-media + NetSuite revenue + GA4 traffic data. There are levers and constraints inside the Client's P&L, product economics, and operations that we can't see from the outside. Below are the questions that β once answered β would either confirm the current recommendations or meaningfully reshape them. Group these into the next working session with the the client team.
A. P&L & unit economics β what's the actual breakeven?
- What's the contribution margin per order after COGS, fulfillment, payment processing, and returns? The dashboard treats anything above 2.0x platform-reported ROAS as "above break-even" by convention, but the Client's actual breakeven could be 2.5x, 3.5x, or higher depending on margin structure.
- What's the target NS-ROAS for sustainability? The 5.09x Q1 '26 figure looks bad against the 9.84x 2023 baseline, but is 5.09x actually unprofitable, or is it acceptable depending on the business stage?
- Are there fixed costs we should factor in? If the Client has team / warehouse / platform costs that scale with volume, the optimization changes from "max ROAS" to "max contribution dollars after fixed costs."
B. Product category margins β should we be optimizing for mix?
- What are the gross margins by product category? Mattresses ($1,200+ ASP) vs. Mattress Toppers vs. Protectors vs. Pillows vs. Sheets β are these similar margins, or is the mattress carrying the business while accessories are loss-leaders / volume plays?
- Are bundled orders more or less profitable than single-SKU? The bundling drop in Feb '26 hurt AOV β but if bundles were below-margin (heavily discounted), the AOV drop might be margin-neutral or positive. Need to know.
- Are any products strategically loss-leaders? If pillows / sheets are intentionally low-margin to win the mattress conversion, the campaign-type ROAS analysis should be re-weighted.
- Mattress size mix matters too β King and Cal-King margins should be materially better than Twin. Is the size mix moving in a favorable direction?
C. Customer LTV & cohort behavior β first-purchase or lifetime?
- What's the typical repeat-purchase rate? Mattresses are infrequent (5β8 year replacement cycle), but accessories should repeat. What's the 12-month / 24-month repurchase rate?
- LTV by acquisition channel β do customers acquired via Brand search behave the same as customers acquired via NB / Meta / Organic? If NB-acquired customers have higher LTV, the NB ROAS comparison understates NB's true value.
- Are we optimizing for first-purchase profitability or lifetime contribution? If LTV-based, the breakeven ROAS for first-purchase NB campaigns can be lower (say 1.5x or even 1.0x) because the second purchase + accessory cross-sell makes the customer profitable.
- Email / SMS revenue contribution β what % of NetSuite revenue comes from owned channels (email/SMS) vs paid acquisition? If owned-channel post-purchase is strong, paid ROAS should be evaluated as a customer-acquisition cost, not a transaction ROI.
D. Promo strategy & discount profitability
- At what discount depth does an order become net-margin negative? Q4 '25 hit 27.8% discount depth on Black Friday β was that still margin-positive, or volume-only?
- Promo cadence β is the current calendar driven by margin economics or competitive pressure? The pull-forward signal (T+2 r=β0.32) suggests deep promos compress the next 2 months. Worth knowing if this is an acceptable trade.
- Affiliate / influencer commissions β at what payout rate is an affiliate transaction profitable vs. a margin drag? These show up in the "(not set)" GA4 attribution leakage.
E. Operational & strategic context β what's driving the decisions we observed?
- Was the Feb '26 bundling stop deliberate, or a side effect (NetSuite labeling change, e-comm site change, SKU restructure)? The "exactly 1.0" attach rate suggests measurement, not behavior β needs the Client-side confirmation.
- What was the rationale for the Q1 '26 NB / PMax pullback? Cash management, target ROAS hit, leadership change, inventory constraints, supply-chain issue, profitability targeting? Different root causes change the recommendation.
- What's the 2026 plan vs. 2025 strategy? Is the Client in growth mode, profitability mode, or hold-position mode? The 70/30 recommendation is calibrated for growth; if the goal is profit-protection, a different mix may be appropriate.
- Inventory & capacity constraints β are there months where the Client can't scale demand because inventory or fulfillment is the bottleneck? This affects whether NB scaling is even safe to pursue.
- Are there recent product launches, brand partnerships, or PR moments that would explain demand changes the data alone can't?
How to use this view: these questions don't replace the recommendations β they sharpen them. The 70/30 NB/Brand split is supported by the data we have today. If the answers to questions A and B show that mattresses carry significantly higher margins than accessories, the recommendation might shift from "70/30 NB/Brand spread evenly across the NB toolkit" to "70/30 NB/Brand weighted toward channels that drive mattress conversion." The shape of the answer changes the shape of the action β that's why these questions matter.
Suggested next step: bring Groups A, B, and E to the kickoff working session with the Client. Groups C (LTV) and D (promo profitability) can be follow-ups once the basics are in place.
Suggested next step: bring Groups A, B, and E to the kickoff working session with the Client. Groups C (LTV) and D (promo profitability) can be follow-ups once the basics are in place.
Overview β NetSuite Revenue Anchor
Monthly NetSuite revenue is our dependent variable. Every other tab in this dashboard exists to explain why revenue deviated from what we'd expect.
The expectation is a blend of three signals, shown below as separate lines so you can see when they disagree:
1. YoY β prior-year same month (full 40-month history, Jan '23 β Apr '26) Β· 2. Trend β trailing 3-month average Β· 3. ROI-implied β this month's total paid spend Γ trailing 6-month rev/spend ratio
1. YoY β prior-year same month (full 40-month history, Jan '23 β Apr '26) Β· 2. Trend β trailing 3-month average Β· 3. ROI-implied β this month's total paid spend Γ trailing 6-month rev/spend ratio
The Quad Anchor β NetSuite Revenue Β· AOV Β· Orders Β· NetSuite ROI
The four output metrics the Client cares about, stacked on a shared time axis. Driver trends below use the same time axis so you can eyeball correlations vertically. Apr 2026 covers days 1β21 only β lighter/dashed bars are projected remainder at current pace (Γ30/21). AOV and NetSuite ROI are ratios and aren't projected. NetSuite ROI = NetSuite revenue Γ· total paid marketing spend (Google+Meta+Bing) β the honest bottom-line return on ad dollars.
π Multi-Year Context β NetSuite Revenue, Total Paid Spend & NS-ROAS (Jan 2023 β Apr 2026)
Bars = monthly NetSuite net revenue. Amber line = total paid spend (Google+Meta+Bing). Teal line (right axis) = NetSuite ROAS. The long-horizon view that anchors the current-state Jan '25βApr '26 focus window below.
What the 40-month view shows:
Β· 2023 baseline β $3.27M NS, 7.88x NS-ROAS. Small paid footprint ($415K).
Β· 2024 peak year β $4.35M NS (+33% YoY), 7.61x NS-ROAS. Paid scaled to $572K.
Β· 2025 plateau β $4.20M NS (β3.6% YoY). Paid grew to $700K, but NS-ROAS dropped to 5.99x β the efficiency leak started here, not in Q1 '26.
Β· Q1 '26 acceleration β paid at ~$188K in 4 months (annualized pace ~$565K+), NS-ROAS fell to 5.09x. Spend +28% YoY while NS revenue β21% YoY.
Β· 2023 baseline β $3.27M NS, 7.88x NS-ROAS. Small paid footprint ($415K).
Β· 2024 peak year β $4.35M NS (+33% YoY), 7.61x NS-ROAS. Paid scaled to $572K.
Β· 2025 plateau β $4.20M NS (β3.6% YoY). Paid grew to $700K, but NS-ROAS dropped to 5.99x β the efficiency leak started here, not in Q1 '26.
Β· Q1 '26 acceleration β paid at ~$188K in 4 months (annualized pace ~$565K+), NS-ROAS fell to 5.09x. Spend +28% YoY while NS revenue β21% YoY.
Monthly NetSuite Revenue β Actuals vs. YoY & Seasonal Expectation (Focus Window: Jan '25 β Apr '26)
Bars = actual NetSuite revenue. Amber = YoY (prior-year same month NetSuite, Jan '24+). Teal dashed = seasonally-adjusted expectation β trailing-6-mo deseasonalized mean Γ this month's historical seasonal factor. This pulls Jan/Mar baselines down and Nov up, correcting the earlier 3-mo-rolling-avg bias that punished post-holiday months.
Driver Trends β Aligned to the Anchor Time Axis
Three headline driver metrics on the same x-axis as the anchor above β the highest-signal levers for explaining NetSuite revenue moves. Full breakout for each driver lives in its dedicated tab.
β Platform Spend and Brand Share cover the full 40 months (Jan '23 β Apr '26). Paid Search share of sessions uses GA4 data and starts Feb '25.
Brand Share of Google+Bing Spend (%)
Platform Spend β Google Β· Meta Β· Bing ($)
Paid Search Share of Sessions (%) Β· GA4, Feb '25+
Driver #5 β Platform Budget Share Β· Does Mix Shift Move NetSuite Revenue?
Core question: when the Client shifts money between Google / Meta / Bing, does NetSuite bottom-line revenue actually respond?
40-month view (Jan '23 β Apr '26): Google has been the dominant platform every month (74-83% of paid). Total annual paid: 2023 $416K β 2024 $572K (+38%) β 2025 $700K (+22%). Bing launched Oct '23, ran ~$67-69K/year in 2024-2025, then shrank to 5% share in Q1 '26. Meta was trivial through 2023 ($21K/year), scaled in 2024 ($81K), held in 2025 ($79K), and is on pace for a record year in 2026 (Q1 '26 Meta spend $32K already = 40% of 2025 full-year).
Two ways to read this tab:
1. Platform share vs NetSuite outcome β compare the monthly platform-share charts to the NetSuite revenue trend on the Revenue Anchor tab. If Meta share rises 10pp in a month and NetSuite revenue doesn't move, the reallocation was neutral.
2. Blended NetSuite ROI β see the ROI trend in the Revenue Anchor's Quad Anchor chart. That's the cleanest view of whether total paid spend is delivering more or less NetSuite revenue per dollar over time.
Per-platform ROAS numbers shown below are Google/Meta/Bing-reported (not NetSuite). They're useful for within-platform trend analysis only β don't add them across platforms (they double-count shared conversions).
40-month view (Jan '23 β Apr '26): Google has been the dominant platform every month (74-83% of paid). Total annual paid: 2023 $416K β 2024 $572K (+38%) β 2025 $700K (+22%). Bing launched Oct '23, ran ~$67-69K/year in 2024-2025, then shrank to 5% share in Q1 '26. Meta was trivial through 2023 ($21K/year), scaled in 2024 ($81K), held in 2025 ($79K), and is on pace for a record year in 2026 (Q1 '26 Meta spend $32K already = 40% of 2025 full-year).
Two ways to read this tab:
1. Platform share vs NetSuite outcome β compare the monthly platform-share charts to the NetSuite revenue trend on the Revenue Anchor tab. If Meta share rises 10pp in a month and NetSuite revenue doesn't move, the reallocation was neutral.
2. Blended NetSuite ROI β see the ROI trend in the Revenue Anchor's Quad Anchor chart. That's the cleanest view of whether total paid spend is delivering more or less NetSuite revenue per dollar over time.
Per-platform ROAS numbers shown below are Google/Meta/Bing-reported (not NetSuite). They're useful for within-platform trend analysis only β don't add them across platforms (they double-count shared conversions).
Google 40-mo Spend
β
Dominant platform
Meta 40-mo Spend
β
β
Bing 40-mo Spend
β
β
Dec '25 Google ROAS drop
-30%
5.53x (Nov) β 3.89x (Dec) Β· explains most of Dec deviation
Platform Spend β Monthly Absolute $
Stacked dollars. Watch where total spend spikes and which platform absorbs the budget.
Platform Share of Spend β % Mix
100% stacked. Looking for mix shifts that pre-date NetSuite revenue changes.
Platform ROAS β Monthly
Platform-reported conv. value Γ· spend. Break-even = 2.0x (dashed line).
Paid-Reported Revenue vs. Implied Non-Paid Revenue
Stacked bar shows Google + Meta + Bing reported conversion value, plus the "implied non-paid" remainder needed to reach NetSuite actual.
Negative implied non-paid (Dec '25) means platform claims exceeded NetSuite total β evidence of cross-platform attribution overlap.
π Correlation with the Revenue Anchor β 40-month Pearson r
Computed across Jan '23 β Apr '26 (n=40). Interpretation guide: |r| β₯ 0.5 strong, 0.3β0.5 moderate, < 0.3 weak. T+0 = same-month; T+1..T+3 = 1β3 month lag.
Same-month spend moves NetSuite revenue β nothing else does
Total paid spend β NS rev: r = +0.69 (T+0) β strongest signal in the tab. When the Client spends more this month, NS revenue rises this month.
β’ Google spend β NS rev: +0.68 (near-identical to total β Google is the whole story)
β’ Meta spend β NS rev: +0.49 (moderate)
β’ Bing spend β NS rev: +0.33 (weak, partly because Bing only started Oct '23)
All lagged correlations drop to zero (T+1 = β0.03, T+2 = 0.00, T+3 = +0.07). The Client's business is direct-response dominant β this month's spend doesn't build future months. Cuts hurt this month, not subsequent months.
Total paid spend β NS rev: r = +0.69 (T+0) β strongest signal in the tab. When the Client spends more this month, NS revenue rises this month.
β’ Google spend β NS rev: +0.68 (near-identical to total β Google is the whole story)
β’ Meta spend β NS rev: +0.49 (moderate)
β’ Bing spend β NS rev: +0.33 (weak, partly because Bing only started Oct '23)
All lagged correlations drop to zero (T+1 = β0.03, T+2 = 0.00, T+3 = +0.07). The Client's business is direct-response dominant β this month's spend doesn't build future months. Cuts hurt this month, not subsequent months.
Platform MIX shifts are basically neutral
Changing the Google/Meta/Bing share mix does NOT correlate with NS revenue in any lag window:
β’ Google share β NS: β0.10 T+0, β0.18 T+3 (no signal)
β’ Meta share β NS: β0.06 T+0 (no signal)
β’ Bing share β NS: +0.27 T+0, +0.33 T+3 (weak, confounded by post-Oct '23 ramp)
Practical read: the platform reallocation debate is a second-order question. The first-order question is always total paid spend β mix is mostly noise against NS outcome. The exception is the Meta-vs-Google ROAS gap (see below) where over-allocating to a lower-ROAS platform drags blended efficiency even if it doesn't hurt NS revenue directly.
Changing the Google/Meta/Bing share mix does NOT correlate with NS revenue in any lag window:
β’ Google share β NS: β0.10 T+0, β0.18 T+3 (no signal)
β’ Meta share β NS: β0.06 T+0 (no signal)
β’ Bing share β NS: +0.27 T+0, +0.33 T+3 (weak, confounded by post-Oct '23 ramp)
Practical read: the platform reallocation debate is a second-order question. The first-order question is always total paid spend β mix is mostly noise against NS outcome. The exception is the Meta-vs-Google ROAS gap (see below) where over-allocating to a lower-ROAS platform drags blended efficiency even if it doesn't hurt NS revenue directly.
Platform ROAS β NS revenue β a modest same-month signal
β’ Google ROAS β NS rev: +0.34 T+0 (moderate β Google-reported ROAS tracks NS partially because Google is most of the business)
β’ Meta ROAS β NS rev: β0.17 T+0 (weak negative β Meta ROAS measures Meta's own attribution, which doesn't overlap cleanly with NS)
β’ Bing ROAS β NS rev: +0.10 T+0, β0.38 T+2 (weak same-month, unreliable lagged β small Bing sample)
Google ROAS is the most honest platform-level proxy for NS-outcome direction. Meta/Bing ROAS β use only for within-platform trend comparison.
β’ Google ROAS β NS rev: +0.34 T+0 (moderate β Google-reported ROAS tracks NS partially because Google is most of the business)
β’ Meta ROAS β NS rev: β0.17 T+0 (weak negative β Meta ROAS measures Meta's own attribution, which doesn't overlap cleanly with NS)
β’ Bing ROAS β NS rev: +0.10 T+0, β0.38 T+2 (weak same-month, unreliable lagged β small Bing sample)
Google ROAS is the most honest platform-level proxy for NS-outcome direction. Meta/Bing ROAS β use only for within-platform trend comparison.
Structural β Meta share growing at lower ROAS (real watch-item)
Multi-year platform share arc (non-seasonal):
β’ Google share 2023β2025 stable 74β83%; Q1 '26 = 78% (unchanged)
β’ Meta share: 5% ('23) β 14% ('24) β 11% ('25) β 17% Q1 '26. Meta 2026 Q1 spend $32K is already 40% of 2025 full-year.
β’ Bing share: grew 12% ('23) β 12% ('24) β 10% ('25) β 5% Q1 '26. Declining sharply.
Q1 '26 Meta ROAS (platform-reported 1.88β4.28x) is structurally below Google (3.78β6.28x). Meta share growth at lower ROAS drags blended efficiency. Since Meta share isn't correlated with NS revenue directly, the real cost is efficiency, not volume β you don't lose sales, but you do pay more per dollar of NS revenue.
Multi-year platform share arc (non-seasonal):
β’ Google share 2023β2025 stable 74β83%; Q1 '26 = 78% (unchanged)
β’ Meta share: 5% ('23) β 14% ('24) β 11% ('25) β 17% Q1 '26. Meta 2026 Q1 spend $32K is already 40% of 2025 full-year.
β’ Bing share: grew 12% ('23) β 12% ('24) β 10% ('25) β 5% Q1 '26. Declining sharply.
Q1 '26 Meta ROAS (platform-reported 1.88β4.28x) is structurally below Google (3.78β6.28x). Meta share growth at lower ROAS drags blended efficiency. Since Meta share isn't correlated with NS revenue directly, the real cost is efficiency, not volume β you don't lose sales, but you do pay more per dollar of NS revenue.
Driver #2 β Remarketing vs Prospecting Spend Mix
Cross-platform split: every campaign classified as Remarketing
(name contains Remarketing / RLSA / RT / Retargeting) or Prospecting (everything else). Covers Google + Meta + Bing.
Lens: how does investment in RMKT vs Prospecting (as % of total paid) correlate with NetSuite bottom-line revenue? Segment-level ROAS figures below are platform-reported (useful for within-segment trend, not as ground truth).
Lens: how does investment in RMKT vs Prospecting (as % of total paid) correlate with NetSuite bottom-line revenue? Segment-level ROAS figures below are platform-reported (useful for within-segment trend, not as ground truth).
RMKT Share of Spend (40-mo)
DTC benchmark: 15β30% Β· declining every year: 6.1% ('23) β 4.6% ('24) β 1.9% ('25)
RMKT Spend (40-mo)
β
vs Prospecting β
RMKT ROAS (40-mo)
β
Never above 0.35x annual avg Β· Display RMKT is the consistent drag
Prospecting ROAS trajectory
2.47x β 4.48x
2023 β 2025 Β· prospecting engine nearly 2Γ more efficient
Monthly Spend β RMKT vs Prospecting
Stacked $. Watch for whether investment in the prospecting engine is what actually moves NetSuite revenue.
RMKT Share of Spend β %
Rarely above 2%. One small spike (FebβMar '25 at 4-5% from Bing). April '26 is the first material push.
ROAS by Segment β RMKT vs Prospecting
Prospecting consistently 3β6x. RMKT mostly 0.00x because Display RMKT spend generates almost no attributed revenue. Break-even at 2.0x (dashed).
π Correlation with the Revenue Anchor β 40-month Pearson r
Computed across Jan '23 β Apr '26 (n=40). |r| β₯ 0.5 strong, 0.3β0.5 moderate, < 0.3 weak.
Prospecting is the NS revenue engine β RMKT isn't
Prospecting spend β NS revenue: r = +0.67 (T+0) β strong same-month driver. Matches the +0.69 for Total paid (prospecting is ~97% of spend).
RMKT spend β NS revenue: r = +0.31 (T+0) β weak. That "correlation" largely comes from RMKT co-moving with prospecting within the same monthly promo cycles, not RMKT causing NS revenue.
Prospecting ROAS (platform-reported) β NS rev: +0.27 T+0 (weak positive). Lagged correlations all β 0 β no delayed effect of either RMKT or Prospecting spend on NS revenue.
Prospecting spend β NS revenue: r = +0.67 (T+0) β strong same-month driver. Matches the +0.69 for Total paid (prospecting is ~97% of spend).
RMKT spend β NS revenue: r = +0.31 (T+0) β weak. That "correlation" largely comes from RMKT co-moving with prospecting within the same monthly promo cycles, not RMKT causing NS revenue.
Prospecting ROAS (platform-reported) β NS rev: +0.27 T+0 (weak positive). Lagged correlations all β 0 β no delayed effect of either RMKT or Prospecting spend on NS revenue.
RMKT share is weakly negatively correlated β but that's structural
RMKT share of spend β NS rev: r = β0.26 (T+0) β weak negative. The mechanism is simple: RMKT share rises most when prospecting is cut (denominator effect), and those same months tend to be lower-revenue (post-BFCM pullbacks, off-promo windows). The signal says "months with higher RMKT share tend to be smaller months" β it's a symptom, not a lever.
Takeaway: you can't "fix" NS revenue by raising RMKT share. You can only fix it by raising RMKT spend in absolute terms alongside prospecting, and that hasn't been tested meaningfully yet (see below).
RMKT share of spend β NS rev: r = β0.26 (T+0) β weak negative. The mechanism is simple: RMKT share rises most when prospecting is cut (denominator effect), and those same months tend to be lower-revenue (post-BFCM pullbacks, off-promo windows). The signal says "months with higher RMKT share tend to be smaller months" β it's a symptom, not a lever.
Takeaway: you can't "fix" NS revenue by raising RMKT share. You can only fix it by raising RMKT spend in absolute terms alongside prospecting, and that hasn't been tested meaningfully yet (see below).
Structural β the 40-month RMKT gap
RMKT has been shrinking every year as prospecting scaled:
β’ 2023: 6.1% RMKT share (peak 10.8%)
β’ 2024: 4.6% avg (peak 9.3%)
β’ 2025: 1.9% avg (peak 5.1%)
β’ Q1 '26: ~1.5% avg Β· Apr jumped to 5.5%
DTC benchmark is 15β30%. Google Display Remarketing ran 0.30x ('23), 0.31x ('24), 0.17x ('25) ROAS annually β a consistent drag.
Caveat on the correlation read above: RMKT has never been run at DTC-benchmark scale, so the "RMKT spend has low NS-correlation" finding is a statistical artifact of the tiny investment β not evidence that RMKT doesn't work at scale.
RMKT has been shrinking every year as prospecting scaled:
β’ 2023: 6.1% RMKT share (peak 10.8%)
β’ 2024: 4.6% avg (peak 9.3%)
β’ 2025: 1.9% avg (peak 5.1%)
β’ Q1 '26: ~1.5% avg Β· Apr jumped to 5.5%
DTC benchmark is 15β30%. Google Display Remarketing ran 0.30x ('23), 0.31x ('24), 0.17x ('25) ROAS annually β a consistent drag.
Caveat on the correlation read above: RMKT has never been run at DTC-benchmark scale, so the "RMKT spend has low NS-correlation" finding is a statistical artifact of the tiny investment β not evidence that RMKT doesn't work at scale.
Apr '26 signal β first real RMKT activation test
Apr '26 RMKT spend jumped to $2,297 (vs $760 baseline) β $8,027 platform-reported revenue at 3.49x ROAS β highest RMKT ROAS in the 40-month dataset.
Driven by Meta EarthDay2026_RT campaigns. This is a seasonal tailwind (Earth Day is big for an organic mattress brand) combined with the Client's first real Meta RT test.
What to watch: if the evergreen Meta RT (non-promo-tied) sustains even 2.0x+ ROAS post-Earth Day, it's evidence the RMKT gap can be closed. That would be the first real change to the 40-month structural story.
Apr '26 RMKT spend jumped to $2,297 (vs $760 baseline) β $8,027 platform-reported revenue at 3.49x ROAS β highest RMKT ROAS in the 40-month dataset.
Driven by Meta EarthDay2026_RT campaigns. This is a seasonal tailwind (Earth Day is big for an organic mattress brand) combined with the Client's first real Meta RT test.
What to watch: if the evergreen Meta RT (non-promo-tied) sustains even 2.0x+ ROAS post-Earth Day, it's evidence the RMKT gap can be closed. That would be the first real change to the 40-month structural story.
Driver #3 β Brand vs Non-Brand Spend Mix
Google + Bing campaigns classified as Brand, Non-Brand, Competitor, or Shopping. Bidirectional search-term leakage adjustment applied: Brand-campaign spend on non-brand-name queries reclassified to NB AND NB-campaign spend on the Client queries reclassified to Brand. Both spend and conversion value adjusted proportionally across full campaign cost. Methodology at bottom. ROAS figures are platform-reported.
Brand ROAS (40-mo avg)
β
Climbed steeply: 3.80x ('23) β 5.54x ('24) β 8.26x ('25)
NB ROAS (40-mo avg)
β
0.94x ('23) β 1.53x ('24) β 2.54x ('25) Β· 2.7Γ improvement, still below comfort
Q1 '26 NB Spend Cut vs 2025 pace
-36%
$61K Q1 '26 = $244K annualized vs $384K 2025 Β· steepest NB pullback in 40-mo history
Brand share regime flip
36% β 52%
2025 avg β Q1 '26 Β· reverses the 3-year NB-scale trend
Monthly Spend β Segment Stack (40 months)
Google + Bing spend by segment, Jan '23 β Apr '26. Notable structural shifts: (1) Bing launched Oct '23 Β· (2) NB scaled steadily through 2024-2025 Β· (3) Q1 '26 NB cut sharp + Shopping launched March '26. Note: Bing has no campaign-level brand/NB classification possibility for all campaigns β what's captured is summed into the Google-based buckets.
Segment Share of Spend β % (40 months)
100% stacked, Jan '23 β Apr '26. Multi-year arc: Brand dominant in 2023 (56%), NB-scaled through 2024-2025 (49% β 36%), then snapped back to 52% Q1 '26.
Segment ROAS β Monthly (40 months)
Brand climbed 3.8x β 5.5x β 8.3x ('23β'25). NB climbed 0.94x β 1.5x β 2.5x. Break-even at 2.0x (dashed) β NB only crossed it in 2025 (and barely).
π Annual totals (Google+Bing, bidirectionally adjusted)
| Year | Brand Spend | Brand ROAS | NB Spend | NB ROAS | Brand Share |
|---|---|---|---|---|---|
| 2023 | $188K | 3.80x | $149K | 0.94x | 56% |
| 2024 | $225K | 5.54x | $235K | 1.53x | 49% |
| 2025 | $216K | 8.26x | $384K | 2.54x | 36% |
| Q1 '26 | $66K | 7.00x | $61K | 2.53x | 52% |
4-year arc (bidirectional reclassification): NB spend scaled 2.6Γ ('23β'25, $149K β $384K) and NB ROAS improved 0.94x β 1.53x β 2.54x β that's the prospecting build-out behind the revenue growth. Brand efficiency compounded too (3.80x β 5.54x β 8.26x β steeper improvement than the campaign-name-only view showed, because branded conversions previously credited to NB campaigns are now correctly attributed to Brand). Brand share compressed naturally as NB scaled (56% β 36%). Q1 '26 reverses this arc β Brand share back to 52%, NB at a $244K annualized pace (vs $384K in 2025, β36% YoY).
Segment Revenue (Google-reported) β Monthly
Google-reported conversion value per segment (NOT NetSuite). NB has historically been the volume driver; Brand is the margin contributor. Useful for within-segment comparison; don't read as bottom-line revenue.
π§ Brand Cannibalization β Is Brand Spend Consuming Organic/Direct?
Hypothesis: if Brand spend is capturing demand that organic/direct would have closed anyway, we'd see total non-CPC traffic volume flat while Brand's share of that traffic grows. That's classic cannibalization β paying for visits you'd get for free. Data: GA4 non-CPC traffic. Pre-Sep '25 is directional only.
β οΈ Q1 '26 β the rising Search IS is a symptom, not a win
Observed: overall paid-search IS rose Jan 46% β Feb 68% β Mar 87% β Apr 85% (vs 25β35% baseline). Looks like dominance.
Reality β the rise is 96β100% Brand-driven: NB Search IS has never crossed 14% in any campaign-month across the entire 40-month record. The IS spike is mechanical: NB spend was cut 36% YoY while Brand held, so the denominator (NB impressions) collapsed and Brand's share looked larger. Absolute impression count went down.
Why this contributes to NS revenue weakness: (1) Cannibalization β Brand search captures people who already know the Client. Brand share of paid is 52% (vs 3-yr 36β56% with bidirectional adjustment) and Brand IS at 85β95% means paying to harvest existing demand that would have converted via organic/direct. Non-CPC Brand traffic share rose 7β10pp Q1 '26 without total traffic rising. (2) Top-of-funnel starvation β NB is the only channel generating new awareness. NB at $20K/mo Q1 '26 (vs $32K/mo 2025 avg) means fewer new shoppers entering the funnel. Consistent with the β33% Jan / β16% Mar / β12% Apr YoY NS revenue gaps. (3) Attribution lag β Brand's 7β10x bidirectionally-corrected ROAS already accounts for branded conversions previously credited to NB. The remaining lag risk: as the brand-aware audience ages out without NB to refill it, even Brand's new high ROAS will start to compress (per the Lag Effects β0.63 T+3 signal).
Action: stop reading the IS rise as positive Β· track NB Search IS independently Β· restore NB investment (or upper-funnel substitutes β Shopping / Meta / Video) before Brand ROAS starts eroding.
Reality β the rise is 96β100% Brand-driven: NB Search IS has never crossed 14% in any campaign-month across the entire 40-month record. The IS spike is mechanical: NB spend was cut 36% YoY while Brand held, so the denominator (NB impressions) collapsed and Brand's share looked larger. Absolute impression count went down.
Why this contributes to NS revenue weakness: (1) Cannibalization β Brand search captures people who already know the Client. Brand share of paid is 52% (vs 3-yr 36β56% with bidirectional adjustment) and Brand IS at 85β95% means paying to harvest existing demand that would have converted via organic/direct. Non-CPC Brand traffic share rose 7β10pp Q1 '26 without total traffic rising. (2) Top-of-funnel starvation β NB is the only channel generating new awareness. NB at $20K/mo Q1 '26 (vs $32K/mo 2025 avg) means fewer new shoppers entering the funnel. Consistent with the β33% Jan / β16% Mar / β12% Apr YoY NS revenue gaps. (3) Attribution lag β Brand's 7β10x bidirectionally-corrected ROAS already accounts for branded conversions previously credited to NB. The remaining lag risk: as the brand-aware audience ages out without NB to refill it, even Brand's new high ROAS will start to compress (per the Lag Effects β0.63 T+3 signal).
Action: stop reading the IS rise as positive Β· track NB Search IS independently Β· restore NB investment (or upper-funnel substitutes β Shopping / Meta / Video) before Brand ROAS starts eroding.
Brand Traffic Share Jump (Q1 '26)
+7-10pp
Sep-Dec '25: ~9% avg Β· Feb-Apr '26: 13-18%
Brand Revenue Share (Q1 '26)
28-33%
vs H2 '25 avg ~28% β attribution rose without total revenue rising
Feb '26 Paid Brand Share Spike
51%
Bidirectionally-adjusted Brand cost $19.9K vs NB $18.9K Β· flipped from 23% in Dec '25
NB Search IS β Demand Left on the Table
~65β75% lost
Competitors capture most NB impressions Β· see below
Brand Traffic % of Non-CPC Sessions β Monthly
Brand session share of non-CPC traffic. Rising without total traffic rising = cannibalization signal.
Brand Revenue % of Non-CPC Revenue β Monthly
Brand revenue share of non-CPC revenue. Same lens as traffic but on dollars.
Google Ads Paid β NB vs Brand Cost (Monthly)
What the Client actually spent on NB vs Brand in Google paid. Watch for periods where Brand share of paid rises while total NetSuite revenue doesn't keep pace β that's the over-investment signal.
π Correlation with the Revenue Anchor β 40-month Pearson r
Computed across Jan '23 β Apr '26 (n=40). |r| β₯ 0.5 strong, 0.3β0.5 moderate, < 0.3 weak.
Both Brand and NB spend correlate with NS revenue β but Brand share is the warning signal
Same-month (T+0), bidirectionally adjusted:
Β· Brand spend β NS: +0.83 (strongest single line item)
Β· NB spend β NS: +0.58 (strong)
Β· Brand share of spend β NS: β0.36 (moderate negative)
Β· NB share of spend β NS: +0.47 (moderate positive)
Reading: higher absolute spend on either segment moves NS revenue. But the mix matters β Brand-heavy months (Brand > NB) tend to have lower NS revenue, and NB-heavy months tend to have higher NS revenue.
Same-month (T+0), bidirectionally adjusted:
Β· Brand spend β NS: +0.83 (strongest single line item)
Β· NB spend β NS: +0.58 (strong)
Β· Brand share of spend β NS: β0.36 (moderate negative)
Β· NB share of spend β NS: +0.47 (moderate positive)
Reading: higher absolute spend on either segment moves NS revenue. But the mix matters β Brand-heavy months (Brand > NB) tend to have lower NS revenue, and NB-heavy months tend to have higher NS revenue.
Brand-share negative signal also persists at T+2
Multi-lag read:
Β· Brand share β NS rev: T+0 β0.36, T+1 β0.14, T+2 β0.22, T+3 +0.05
Β· NB share β NS rev: T+0 +0.47, T+1 +0.19, T+2 +0.27, T+3 β0.01
The NB share β NS-revenue link isn't just same-month β there's a weak 1β2 month carry. And the Lag Effects tab shows a stronger separate signal: Brand share β NB ROAS T+3 = β0.63 (strong negative). Taken together: Brand-heavy mix both underperforms this month AND compresses NB efficiency 2β3 months later.
Multi-lag read:
Β· Brand share β NS rev: T+0 β0.36, T+1 β0.14, T+2 β0.22, T+3 +0.05
Β· NB share β NS rev: T+0 +0.47, T+1 +0.19, T+2 +0.27, T+3 β0.01
The NB share β NS-revenue link isn't just same-month β there's a weak 1β2 month carry. And the Lag Effects tab shows a stronger separate signal: Brand share β NB ROAS T+3 = β0.63 (strong negative). Taken together: Brand-heavy mix both underperforms this month AND compresses NB efficiency 2β3 months later.
Q1 '26 FebβApr β deepest NB pullback in 40-month history
NB spend (Google+Bing, bidirectionally adjusted) fell from $384K/year (2025) to $244K Q1 '26 annualized pace β that's ~β36% YoY. This is still materially beyond seasonal Q1 pullback norms: historical Q1 NB contraction vs the Q4 baseline was typically 30β50%, but here it follows after the bidirectional reclassification revealed NB was already running at lower ROAS than campaign-name reporting suggested.
Brand share of paid search jumped from 36% (2025 avg) to 52% (Q1 '26), reversing a 3-year NB-scale arc (56% β 49% β 36% β 52%). At current Brand share the Lag model predicts continued NB ROAS compression through MayβJul '26.
NB ROAS held at 2.5x (above break-even but barely) β the risk is sustained underperformance + volume contraction.
NB spend (Google+Bing, bidirectionally adjusted) fell from $384K/year (2025) to $244K Q1 '26 annualized pace β that's ~β36% YoY. This is still materially beyond seasonal Q1 pullback norms: historical Q1 NB contraction vs the Q4 baseline was typically 30β50%, but here it follows after the bidirectional reclassification revealed NB was already running at lower ROAS than campaign-name reporting suggested.
Brand share of paid search jumped from 36% (2025 avg) to 52% (Q1 '26), reversing a 3-year NB-scale arc (56% β 49% β 36% β 52%). At current Brand share the Lag model predicts continued NB ROAS compression through MayβJul '26.
NB ROAS held at 2.5x (above break-even but barely) β the risk is sustained underperformance + volume contraction.
3-year efficiency compounding (the positive context for sizing)
Both Brand and NB ROAS improved steadily as paid media scaled 2023 β 2025:
Β· Brand ROAS: 3.80x ('23) β 5.54x ('24) β 8.26x ('25)
Β· NB ROAS: 0.94x ('23) β 1.53x ('24) β 2.54x ('25)
NB's 2.7Γ efficiency gain across 3 years is the main argument against cutting NB: the Client's own data shows NB campaigns scale toward efficiency over time, even though they're still below the Brand efficiency level (because Brand campaigns harvest existing intent β that's expected). NB is the demand-creation engine; Brand is the harvester.
For the prescriptive Brand-share guardrail and the rest of the action plan, see the π― Recommendations tab.
Both Brand and NB ROAS improved steadily as paid media scaled 2023 β 2025:
Β· Brand ROAS: 3.80x ('23) β 5.54x ('24) β 8.26x ('25)
Β· NB ROAS: 0.94x ('23) β 1.53x ('24) β 2.54x ('25)
NB's 2.7Γ efficiency gain across 3 years is the main argument against cutting NB: the Client's own data shows NB campaigns scale toward efficiency over time, even though they're still below the Brand efficiency level (because Brand campaigns harvest existing intent β that's expected). NB is the demand-creation engine; Brand is the harvester.
For the prescriptive Brand-share guardrail and the rest of the action plan, see the π― Recommendations tab.
βοΈ Methodology β bidirectional Brand/NB reclassification (click to expand)
The problem: campaign-name classification is unreliable in two directions:
Β· Brand campaigns serve some non-brand-name queries (broad/phrase match catching avocado, dreamcloud, birch, etc.) β these should count as NB
Β· NB campaigns serve some the Client queries (PMax pulling branded searches, broad keywords matching brand) β these should count as Brand
The data: 4 search-term reports (Jan '23 β Apr '26):
Β· Branded campaigns serving branded queries (true Brand)
Β· Branded campaigns serving non-branded queries (Brand β NB leak)
Β· Non-branded campaigns serving branded queries (NB β Brand leak)
Β· Non-branded campaigns serving non-branded queries (true NB)
Methodology: the search-term reports cover 80β95% of campaign spend on average (low-volume queries get aggregated into "Other" by Google). Rather than treating the reports as the universe, we use them as a representative sample to compute monthly leak rates in both directions, then apply those rates proportionally to the full known campaign cost (and conv. value).
Β· Brand β NB leak: 2023 23% Β· 2024 21% Β· 2025 21% Β· 2026 YTD 21% β structural
Β· NB β Brand leak: 2023 13% Β· 2024 16% Β· 2025 9% Β· 2026 YTD 6% β declining (negative-keyword hygiene improving)
Bing remains classified by campaign name only (no Bing search-term reports available). All other charts and KPIs on this tab use bidirectionally-adjusted Brand and NB buckets for spend and conv value.
Β· Brand campaigns serve some non-brand-name queries (broad/phrase match catching avocado, dreamcloud, birch, etc.) β these should count as NB
Β· NB campaigns serve some the Client queries (PMax pulling branded searches, broad keywords matching brand) β these should count as Brand
The data: 4 search-term reports (Jan '23 β Apr '26):
Β· Branded campaigns serving branded queries (true Brand)
Β· Branded campaigns serving non-branded queries (Brand β NB leak)
Β· Non-branded campaigns serving branded queries (NB β Brand leak)
Β· Non-branded campaigns serving non-branded queries (true NB)
Methodology: the search-term reports cover 80β95% of campaign spend on average (low-volume queries get aggregated into "Other" by Google). Rather than treating the reports as the universe, we use them as a representative sample to compute monthly leak rates in both directions, then apply those rates proportionally to the full known campaign cost (and conv. value).
nb_in_brand_rate = NB_query_cost_in_Brand_camp / (Brand_query + NB_query in Brand camp)
b_in_nb_rate = Brand_query_cost_in_NB_camp / (Brand_query + NB_query in NB camp)
Adjusted Brand cost = Brand_camp_cost Γ (1 - nb_in_brand_rate)
+ NB_camp_cost Γ b_in_nb_rate
Adjusted NB cost = NB_camp_cost Γ (1 - b_in_nb_rate)
+ Brand_camp_cost Γ nb_in_brand_rate
(Same logic for conv. value)
Annual leak rates (% of campaign-name spend):Β· Brand β NB leak: 2023 23% Β· 2024 21% Β· 2025 21% Β· 2026 YTD 21% β structural
Β· NB β Brand leak: 2023 13% Β· 2024 16% Β· 2025 9% Β· 2026 YTD 6% β declining (negative-keyword hygiene improving)
Bing remains classified by campaign name only (no Bing search-term reports available). All other charts and KPIs on this tab use bidirectionally-adjusted Brand and NB buckets for spend and conv value.
Driver #4 β Campaign Type Share Β· Does More PMax (or Shopping, Meta, etc.) Move NetSuite Revenue?
Core question: does investing more in any single campaign type correlate with NetSuite bottom-line revenue rising?
How to read this tab:
1. Identify a month where a campaign type's spend or share moved meaningfully (e.g. PMax 53% β 23% Jan '26 β Apr '26, Shopping 0% β 19% in the same window).
2. Cross-check NetSuite revenue on the Revenue Anchor tab for that same month. Did it move with the shift?
3. If NetSuite revenue didn't respond, the reallocation was neutral to bottom-line (even if platform-reported ROAS showed improvement).
Types tracked: Search (Google+Bing), Shopping, PMax, Display, Video, Demand Gen, Meta (all objectives combined β Sales, Traffic, Awareness).
Per-type ROAS numbers are platform-reported. PMax especially reads inflated because it conflates Shopping inventory with display attribution. Use within-type trends only, not absolute comparisons.
How to read this tab:
1. Identify a month where a campaign type's spend or share moved meaningfully (e.g. PMax 53% β 23% Jan '26 β Apr '26, Shopping 0% β 19% in the same window).
2. Cross-check NetSuite revenue on the Revenue Anchor tab for that same month. Did it move with the shift?
3. If NetSuite revenue didn't respond, the reallocation was neutral to bottom-line (even if platform-reported ROAS showed improvement).
Types tracked: Search (Google+Bing), Shopping, PMax, Display, Video, Demand Gen, Meta (all objectives combined β Sales, Traffic, Awareness).
Per-type ROAS numbers are platform-reported. PMax especially reads inflated because it conflates Shopping inventory with display attribution. Use within-type trends only, not absolute comparisons.
PMax Share Collapse
53% β 23%
Jan '26 β Apr '26 Β· -30pp shift
Shopping β New
0% β 19%
Launched Mar '26 Β· ROAS 1.25x β 3.39x
Meta Share of Paid (Q1 '26)
17%
2025 avg 11% β Q1 '26 17% Β· growing
Search ROAS (Q1 '26 avg)
5.83x
Best-efficiency type Β· 5.18x/6.28x/6.04x Feb-Apr
π Campaign-type mix evolution (4-year view)
| Type | 2023 | 2024 | 2025 | Q1 '26 | Arc |
|---|---|---|---|---|---|
| Search (G+B) | $286K Β· 69% | $327K Β· 57% | $328K Β· 47% | $68K Β· 36% | Stable $ absolute, shrinking share as other types grew |
| PMax | $83K Β· 20% | $137K Β· 24% | $280K Β· 40% | $71K Β· 37% | 2.1Γ growth '23β'24, then 2Γ '24β'25 β PMax was the big spend story of 2025 |
| Meta (all) | $21K Β· 5% | $81K Β· 14% | $79K Β· 11% | $32K Β· 17% | 4Γ jump '23β'24, plateaued, rising share in Q1 '26. ROAS 4.83x ('23) β 1.52x ('24) β 3.95x ('25) β 3.16x (Q1 '26) |
| Display | $25K Β· 6% | $26K Β· 5% | $13K Β· 2% | $2K Β· 1% | Steady wind-down |
| Shopping | β | β | β | $11K Β· 6% | Launched Mar '26 Β· scaling fast |
| Video / Demand Gen | β | β | β | $5K Β· 2% | Added Q1 '26 as upper-funnel test |
Key pattern: 2023 the Client was essentially a Search+PMax+Display account (~95% of paid). 2024 Meta scaled up ($21K β $81K). 2025 PMax became the biggest single bucket (40% share). Q1 '26 the account diversified into Shopping/Video while cutting PMax heavily β the most varied campaign-type mix in the Client's 4-year history.
Spend by Campaign Type β Monthly (40 months)
Stacked $, Jan '23 β Apr '26. The 4-year arc: Search dominance β PMax scaling β Q1 '26 diversification with Shopping/Meta/Video additions.
Share of Spend β 100% Stacked
Structural shift in Q1 '26 is most visible here.
ROAS by Campaign Type
Break-even at 2.0x (dashed). Search & PMax lead; Display is the drag; Meta volatile year-over-year.
π Correlation with the Revenue Anchor β 40-month Pearson r
Computed across Jan '23 β Apr '26 (n=40). |r| β₯ 0.5 strong, 0.3β0.5 moderate, < 0.3 weak.
Spend by campaign type vs NS revenue β same-month only
Same-month (T+0) correlations:
Β· Search spend β NS rev: +0.67 (strong β the workhorse channel)
Β· Meta spend β NS rev: +0.49 (moderate β Meta has real but noisier pull on NS revenue)
Β· PMax spend β NS rev: +0.47 (moderate β PMax spend tracks NS revenue but less tightly than Search)
Β· Display spend β NS rev: +0.41 (moderate, mostly driven by Nov/Dec co-movement)
Β· Video spend β NS rev: β0.20 (small sample, Q1 '26 only)
All lag correlations β 0 for every campaign type. No campaign type shows a 1β3 month delayed effect on NS revenue. Direct-response-dominant business, confirmed at campaign-type level.
Same-month (T+0) correlations:
Β· Search spend β NS rev: +0.67 (strong β the workhorse channel)
Β· Meta spend β NS rev: +0.49 (moderate β Meta has real but noisier pull on NS revenue)
Β· PMax spend β NS rev: +0.47 (moderate β PMax spend tracks NS revenue but less tightly than Search)
Β· Display spend β NS rev: +0.41 (moderate, mostly driven by Nov/Dec co-movement)
Β· Video spend β NS rev: β0.20 (small sample, Q1 '26 only)
All lag correlations β 0 for every campaign type. No campaign type shows a 1β3 month delayed effect on NS revenue. Direct-response-dominant business, confirmed at campaign-type level.
Search is the reliable anchor; PMax is the 2025 growth lever
Search spend has been the most stable campaign-type driver: ~$327K/year in both 2024 and 2025, T+0 correlation +0.67. It's also the most efficient β 40-mo avg Search ROAS β 4β6x.
PMax is the big 2024 β 2025 spend lever: $137K β $280K (+$144K, 2Γ). T+0 correlation +0.47 β moderate, meaning PMax scaling did drive same-month NS revenue but less per dollar than Search. That's expected β PMax mixes Brand+NB+catalog targeting with lower-intent placements.
Q1 '26 PMax pullback ($25K Jan β $10K Apr, β61%) is the core "NB volume cut" finding. Combined with the Campaign-type mix shift (37% PMax β Shopping + Meta + Video), the question is whether the replacement mix holds the PMax-driven demand.
Search spend has been the most stable campaign-type driver: ~$327K/year in both 2024 and 2025, T+0 correlation +0.67. It's also the most efficient β 40-mo avg Search ROAS β 4β6x.
PMax is the big 2024 β 2025 spend lever: $137K β $280K (+$144K, 2Γ). T+0 correlation +0.47 β moderate, meaning PMax scaling did drive same-month NS revenue but less per dollar than Search. That's expected β PMax mixes Brand+NB+catalog targeting with lower-intent placements.
Q1 '26 PMax pullback ($25K Jan β $10K Apr, β61%) is the core "NB volume cut" finding. Combined with the Campaign-type mix shift (37% PMax β Shopping + Meta + Video), the question is whether the replacement mix holds the PMax-driven demand.
Q1 '26 restructuring β what the 40-mo record shows
Campaign-type diversification in Q1 '26 is unprecedented in the Client's history. 2023 was Search+PMax+Display (~95%); 2024 added Meta at scale; 2025 was PMax-heavy; Q1 '26 has the most varied mix ever:
Β· PMax $71K (37%) β PMax contraction from $80K/mo BFCM peak
Β· Search $68K (36%) β stable
Β· Meta $32K (17%) β growing share
Β· Shopping $11K (6%) β new Mar '26
Β· Video $5K (2%) β new Q1 '26 test
Risk read: the T+0 correlations all reflect what has already run. Shopping and Video are new and have no 40-month track record. Meta T+0 = +0.49 is the strongest single-type non-Google line β scaling Meta is the one big lever that has 4-year support, though its ROAS has varied wildly year-over-year (1.52x in 2024 β 3.95x in 2025).
Campaign-type diversification in Q1 '26 is unprecedented in the Client's history. 2023 was Search+PMax+Display (~95%); 2024 added Meta at scale; 2025 was PMax-heavy; Q1 '26 has the most varied mix ever:
Β· PMax $71K (37%) β PMax contraction from $80K/mo BFCM peak
Β· Search $68K (36%) β stable
Β· Meta $32K (17%) β growing share
Β· Shopping $11K (6%) β new Mar '26
Β· Video $5K (2%) β new Q1 '26 test
Risk read: the T+0 correlations all reflect what has already run. Shopping and Video are new and have no 40-month track record. Meta T+0 = +0.49 is the strongest single-type non-Google line β scaling Meta is the one big lever that has 4-year support, though its ROAS has varied wildly year-over-year (1.52x in 2024 β 3.95x in 2025).
Within-type efficiency is still at or above historical highs
Despite the Q1 '26 volume compression, per-campaign-type ROAS has not weakened:
Β· Search ROAS Q1 '26: 5.18β6.28x (best-efficiency type, above every prior Q1)
Β· PMax ROAS Q1 '26: holds 3.5β4.5x (above historical PMax avg)
Β· Meta ROAS Q1 '26: 3.16x blended (between 2024's 1.52x low and 2025's 3.95x high β neither a win nor a disaster)
The account performs. The question is purely whether the volume cut was correctly sized for the stage of business growth. With NB/PMax share down from 40% ('25) to 37% and absolute $ down ~2Γ vs 2025 pace, the risk is forgone volume, not inefficient spend.
Despite the Q1 '26 volume compression, per-campaign-type ROAS has not weakened:
Β· Search ROAS Q1 '26: 5.18β6.28x (best-efficiency type, above every prior Q1)
Β· PMax ROAS Q1 '26: holds 3.5β4.5x (above historical PMax avg)
Β· Meta ROAS Q1 '26: 3.16x blended (between 2024's 1.52x low and 2025's 3.95x high β neither a win nor a disaster)
The account performs. The question is purely whether the volume cut was correctly sized for the stage of business growth. With NB/PMax share down from 40% ('25) to 37% and absolute $ down ~2Γ vs 2025 pace, the risk is forgone volume, not inefficient spend.
Driver #1 β Traffic Source Mix
GA4 sessions by channel (Feb '25 β Apr '26; GA4 property started Feb 21 '25).
Lens: how does session-mix investment correlate with NetSuite bottom-line revenue? Use session volume + engagement/bounce as the honest traffic-quality signals. GA4 per-channel revenue is NOT used for ground-truth claims because the Client has significant "(not set)" attribution leakage. For revenue truth always cross-reference NetSuite total + platform-reported splits.
Lens: how does session-mix investment correlate with NetSuite bottom-line revenue? Use session volume + engagement/bounce as the honest traffic-quality signals. GA4 per-channel revenue is NOT used for ground-truth claims because the Client has significant "(not set)" attribution leakage. For revenue truth always cross-reference NetSuite total + platform-reported splits.
Paid Search Share Collapse
48% β 27%
Dec '25 β Apr '26 Β· sessions 17.7K β 6.1K
Organic Search Share
Flat at 7β10% throughout Β· structurally under-developed
Direct Share (stable)
β
13β21% Β· brand recall strength signal
Affiliate Share (stable)
β
7β14% Β· reliable evergreen contributor
Monthly Sessions by Channel β Stacked
Absolute sessions. Watch total volume shrinking in 2026 alongside paid cuts.
Channel Share β 100% Stacked
Relative mix. Paid Search dominance declining, Direct/Organic proportionally rising (despite falling in absolute terms).
Engagement Time β Seconds per Session (weighted avg)
Quality gut-check. Falling engagement = thinner traffic even at same volume.
Bounce Rate by Channel
% of sessions with no engagement. Rising bounce rate despite steady spend = traffic quality deterioration.
π Correlation with the Revenue Anchor β Pearson r (15-month GA4 window)
GA4 property started Feb '25 so traffic-source correlations only cover Feb '25 β Apr '26 (n=15). Sample size is small β treat these as directional, not conclusive. |r| β₯ 0.5 is meaningful at n=15 but not statistically significant without more data.
Channel share β NS revenue (T+0)
Β· SMS share β NS rev: +0.57 (strongest β SMS growth months line up with revenue months, likely a post-BFCM retention pattern)
Β· Paid Search share β NS rev: +0.44 (moderate positive)
Β· Affiliate share β NS rev: +0.12 (weak)
Β· Email share β NS rev: β0.04 (none)
Β· Organic Search share β NS rev: β0.17 (weak negative β organic share rises in lower-revenue months as paid shrinks)
Β· Direct share β NS rev: β0.35 (moderate negative β same dynamic)
Β· Other share β NS rev: β0.45 (moderate negative β "other" tends to be noise/miscategorized)
Read: Paid Search and SMS are the two channels where rising share co-moves with rising NS revenue. Organic and Direct move inversely because they're the denominator when paid is cut.
Β· SMS share β NS rev: +0.57 (strongest β SMS growth months line up with revenue months, likely a post-BFCM retention pattern)
Β· Paid Search share β NS rev: +0.44 (moderate positive)
Β· Affiliate share β NS rev: +0.12 (weak)
Β· Email share β NS rev: β0.04 (none)
Β· Organic Search share β NS rev: β0.17 (weak negative β organic share rises in lower-revenue months as paid shrinks)
Β· Direct share β NS rev: β0.35 (moderate negative β same dynamic)
Β· Other share β NS rev: β0.45 (moderate negative β "other" tends to be noise/miscategorized)
Read: Paid Search and SMS are the two channels where rising share co-moves with rising NS revenue. Organic and Direct move inversely because they're the denominator when paid is cut.
Lagged channel-share signals (weak but interesting)
Β· Paid Social share β NS rev T+1: β0.43 β interesting inverse lag. High paid-social share months tend to precede lower-revenue months. Could be a Meta-heavy month signal (Meta-heavy typically weaker NS), but n=15 is small.
Β· Unattributed share β NS rev T+2: +0.65 β GA4 attribution artifact. Unattributed rises in promo months where cross-device/cross-platform conversion paths increase.
Β· Other share β NS rev T+3: β0.61 β weak-quality traffic (referrer noise) is inversely related to future NS revenue.
Low confidence on all of these β the 15-month window is barely enough for a T+3 correlation to register, let alone be trusted.
Β· Paid Social share β NS rev T+1: β0.43 β interesting inverse lag. High paid-social share months tend to precede lower-revenue months. Could be a Meta-heavy month signal (Meta-heavy typically weaker NS), but n=15 is small.
Β· Unattributed share β NS rev T+2: +0.65 β GA4 attribution artifact. Unattributed rises in promo months where cross-device/cross-platform conversion paths increase.
Β· Other share β NS rev T+3: β0.61 β weak-quality traffic (referrer noise) is inversely related to future NS revenue.
Low confidence on all of these β the 15-month window is barely enough for a T+3 correlation to register, let alone be trusted.
Structural β Organic Search is under-developed
Organic Search at 7β10% sessions across the 15-month GA4 window is flat. In DTC mattress accounts, Organic Search typically grows to 15β25% as brand awareness compounds.
the Client has 3+ years of compounding brand activity (the 40-month paid record and the $14M NS revenue base) β but Organic Search hasn't moved. This suggests one of: (1) SEO under-investment, (2) brand queries cannibalized by paid search (see Brand vs NB tab), or (3) competitive organic SERP (Avocado, Birch, Saatva dominate generic mattress SERPs).
Caveat: organic share's correlation with NS revenue is slightly negative (β0.17) only because it rises when paid shrinks. In absolute terms, growing organic traffic is a lever that doesn't show up in this correlation because it hasn't been pulled.
Organic Search at 7β10% sessions across the 15-month GA4 window is flat. In DTC mattress accounts, Organic Search typically grows to 15β25% as brand awareness compounds.
the Client has 3+ years of compounding brand activity (the 40-month paid record and the $14M NS revenue base) β but Organic Search hasn't moved. This suggests one of: (1) SEO under-investment, (2) brand queries cannibalized by paid search (see Brand vs NB tab), or (3) competitive organic SERP (Avocado, Birch, Saatva dominate generic mattress SERPs).
Caveat: organic share's correlation with NS revenue is slightly negative (β0.17) only because it rises when paid shrinks. In absolute terms, growing organic traffic is a lever that doesn't show up in this correlation because it hasn't been pulled.
Data integrity β Paid Social UTM gap
Paid Social sessions read as 0 through most of 2025 because UTM tagging was inconsistent until Dec '25 (5,734 Paid Social sessions). Pre-Dec Meta paid traffic is in Direct, Referral, or (not set).
Use platform-reported Meta spend (Driver #5 / Platform Mix) as source of truth for Meta volume β the GA4 Meta line is only reliable from Dec '25 forward. This is why the Paid Social T+1 β0.43 signal above is low-confidence.
Paid Social sessions read as 0 through most of 2025 because UTM tagging was inconsistent until Dec '25 (5,734 Paid Social sessions). Pre-Dec Meta paid traffic is in Direct, Referral, or (not set).
Use platform-reported Meta spend (Driver #5 / Platform Mix) as source of truth for Meta volume β the GA4 Meta line is only reliable from Dec '25 forward. This is why the Paid Social T+1 β0.43 signal above is low-confidence.
Seasonality β How Far Off Is Each Month From Expected?
Purpose: apply a seasonal adjustment to the Client's monthly NetSuite revenue expectation so the "is this month good or bad" judgment isn't distorted by seasonality (e.g. Jan always weaker than Nov). Seasonal factor comes from the Client's own historical Google ROAS pattern (2019β2024) β the only multi-year seasonality proxy we have. Results tell you how far off each month came in vs. what the Client would normally do for that month of the year.
Jan '26 Discount Pullback
-7.9pp
Discount 15.8% vs Jan '25's 23.7% β lean-promo choice
Apr '26 Discount Pullback
-4.1pp
19.7% vs Apr '25's 23.8% (Earth Day lighter)
Peak Seasonal Factor (Nov)
1.55Γ
Nov is 55% above annual norm historically
Trough Seasonal Factor (Mar)
0.83Γ
Mar is 17% below annual norm historically
Google ROAS β Multi-Year Monthly Trend (2019β2026)
Year-by-year Google ROAS with the 5-year historical seasonal average (dashed). Useful only for deriving the seasonal shape β we use this pattern as a proxy to build the NetSuite-revenue seasonal expectation. NOT the primary performance signal; that's NetSuite revenue YoY on the Revenue Anchor.
Seasonal ROAS Pattern β Month-of-Year
Avg Google ROAS by calendar month (2019β2024 pooled, 6 years each). Nov peaks, Jan/Mar trough. This is the baseline every monthly reading should be compared against.
Seasonal Spend Pattern β Month-of-Year
the Client's historical spend concentrates in Nov ($35K avg) and Dec ($32.5K avg). Q1 is always the leanest.
Discount Depth by Month β Promo Periods Marked
% of NetSuite gross revenue given as discount. Shaded columns = known promo periods. Shows when NetSuite revenue lift came from deeper discounting vs. ad-driven demand.
Same-Month YoY Comparison β NetSuite Gross Revenue, Spend, Efficiency
NetSuite now covers Jan '23 β Apr '26 (40 months) so every current month has a YoY comparable. Google YoY covers Dec (Dec '23 β '24 β '25) since Google data goes back to 2019.
| Month | NetSuite Gross '25 | NetSuite Gross '26 | Gross YoY | Discount % '25 | Discount % '26 | Google ROAS '25 | Google ROAS '26 |
|---|
Historical December Comparison β Google-Only (3 years)
| December | Google Spend | Google Conv. Value | Google ROAS | YoY ROAS Ξ |
|---|---|---|---|---|
| Dec '23 | $46,358 | $83,360 | 1.80x | β |
| Dec '24 | $69,642 | $225,032 | 3.23x | +79% |
| Dec '25 | $73,642 | $286,495 | 3.89x | +20% |
Dec '25 Google ROAS was up YoY on a platform-reported basis (3.89x vs Dec '24 3.23x vs Dec '23 1.80x). That's useful for confirming the seasonal-factor model isn't crazy β Dec is normally weaker than Nov, and the multi-year pattern confirms it. For the NetSuite bottom-line read of Dec '25, see the Revenue Anchor tab; platform-reported Google ROAS is not the primary judgment metric.
π How This Feeds Back Into the Revenue Anchor
Dec '25 context
The -27% deviation vs. a trailing-3-month-average baseline was largely a methodology artifact β that baseline was pulled up by Oct-Nov, which always over-perform pre-BFCM. The seasonal-adjusted expectation (Revenue Anchor tab) is the honest read.
Auction pressure in Dec (Lost IS rank +8.6pp) is a real recurring pattern, not a one-off event.
The -27% deviation vs. a trailing-3-month-average baseline was largely a methodology artifact β that baseline was pulled up by Oct-Nov, which always over-perform pre-BFCM. The seasonal-adjusted expectation (Revenue Anchor tab) is the honest read.
Auction pressure in Dec (Lost IS rank +8.6pp) is a real recurring pattern, not a one-off event.
Jan '26 context
Jan '26 NetSuite revenue was -33.5% YoY β the biggest YoY miss of any comparable month. Partially explained by:
Β· Discount pullback -7.9pp (lighter promo = lower top-line, potentially higher margin)
Β· Net of the discount effect, the residual YoY gap is ~-20-25% β still meaningful.
Real demand-side softness concentrated in Jan '26 is the open question Seasonality alone doesn't explain. See Revenue Anchor's driver correlations.
Jan '26 NetSuite revenue was -33.5% YoY β the biggest YoY miss of any comparable month. Partially explained by:
Β· Discount pullback -7.9pp (lighter promo = lower top-line, potentially higher margin)
Β· Net of the discount effect, the residual YoY gap is ~-20-25% β still meaningful.
Real demand-side softness concentrated in Jan '26 is the open question Seasonality alone doesn't explain. See Revenue Anchor's driver correlations.
Promo-month interpretation
"Over-performance" months on the anchor (+43% May, Nov +15% vs non-seasonal trend) are promo months β Memorial Day, BFCM. Seasonally, they're expected to outperform. The useful question isn't "did we beat expectation?" but "did we beat our own historical promo performance?"
Nov '25 delivered $509K β strong in absolute terms, but the NetSuite-side Q1 '26 follow-through suggests Nov might have pulled forward demand rather than built sustained pipeline.
"Over-performance" months on the anchor (+43% May, Nov +15% vs non-seasonal trend) are promo months β Memorial Day, BFCM. Seasonally, they're expected to outperform. The useful question isn't "did we beat expectation?" but "did we beat our own historical promo performance?"
Nov '25 delivered $509K β strong in absolute terms, but the NetSuite-side Q1 '26 follow-through suggests Nov might have pulled forward demand rather than built sustained pipeline.
Strategic read β margin vs volume tradeoff
Top-line NetSuite revenue is softening YoY (Jan-Apr '26 total -12.7% YoY) because:
(1) The Client chose lighter promos in Jan and Apr '26 β margin-preserving choice
(2) Bundling / multi-product orders stopped Feb '26 β this is the operational issue to fix first
(3) Brand spend share rose while total traffic stayed flat β cannibalization signal
Seasonality is not the story. These three are. See Product Sales, Brand vs NB, and Revenue Anchor tabs.
Top-line NetSuite revenue is softening YoY (Jan-Apr '26 total -12.7% YoY) because:
(1) The Client chose lighter promos in Jan and Apr '26 β margin-preserving choice
(2) Bundling / multi-product orders stopped Feb '26 β this is the operational issue to fix first
(3) Brand spend share rose while total traffic stayed flat β cannibalization signal
Seasonality is not the story. These three are. See Product Sales, Brand vs NB, and Revenue Anchor tabs.
Cross-Month Dependencies β What Happens Next?
Tests two specific hypotheses:
(1) Brand over-indexing: if we run high Brand % of spend for several months, do subsequent months struggle?
(2) Spend reductions: when we cut spend to improve efficiency, how much does it cost us in following months?
Uses 40 months of paid-media history and 40 months of NetSuite revenue (Jan '23 β Apr '26). With the full 40-month NetSuite dataset, longer-lag revenue correlations (T+1, T+2, T+3) now have meaningful statistical power.
Note: Jan 2025 β Apr 2026 Brand/NB spend reflects the true-brand reclassification β broad-match search terms inside Brand campaigns that didn't contain "[brand name]" were moved to NB (see Brand vs NB tab). Pre-2025 values are unadjusted (no search-term data available for that period).
(1) Brand over-indexing: if we run high Brand % of spend for several months, do subsequent months struggle?
(2) Spend reductions: when we cut spend to improve efficiency, how much does it cost us in following months?
Uses 40 months of paid-media history and 40 months of NetSuite revenue (Jan '23 β Apr '26). With the full 40-month NetSuite dataset, longer-lag revenue correlations (T+1, T+2, T+3) now have meaningful statistical power.
Note: Jan 2025 β Apr 2026 Brand/NB spend reflects the true-brand reclassification β broad-match search terms inside Brand campaigns that didn't contain "[brand name]" were moved to NB (see Brand vs NB tab). Pre-2025 values are unadjusted (no search-term data available for that period).
π How to read the numbers on this tab (plain-English)
Correlation coefficient (r): a number between β1 and +1 that tells you whether two things move together.
- r = +1 β perfect positive relationship. When A goes up, B goes up by a proportional amount, every time.
- r = 0 β no relationship. A and B move independently.
- r = β1 β perfect negative relationship. When A goes up, B goes down every time.
What counts as "meaningful" with real-world business data:
- |r| β₯ 0.5 (Strong): a clear pattern. Worth acting on.
- 0.3 β€ |r| < 0.5 (Moderate): a suggestive pattern. Worth watching.
- |r| < 0.3 (Weak): likely noise. Don't anchor decisions on it.
Lag notation (T, T+1, T+2, T+3): T is "this month," T+1 is "one month later," etc. If we're testing "does high Brand share this month predict worse NB efficiency next month?" we'd look at Brand share (T) correlated with NB ROAS (T+1).
Actual calculation: Pearson correlation β for every pair of months we have, we compute how much the two metrics moved in lockstep, then average across all the month pairs.
The big caveat β correlation is not causation: a strong correlation tells us two things move together but not why. For example: if we find high Brand share months correlate with weak subsequent NB ROAS, it could mean (a) brand spend actively cannibalizes NB, OR (b) whenever NB is struggling we shift to Brand, making NB look worse after β opposite causality. Use the correlations as prompts for deeper investigation, not as proof.
Sample-size note: NetSuite revenue now covers 40 months (Jan '23 β Apr '26). At n=40, Pearson correlations are reasonably reliable for T+0 through T+3 lags.
Brand share (T) β Brand ROAS (T+3)
r = -0.61
Strong negative Β· Brand-heavy months predict Brand efficiency decline 3 months later (n=37)
Brand share (T) β NB ROAS (T+3)
r = -0.63
Strong negative Β· NB compresses too (n=37)
Same-month spend β NetSuite revenue
r = +0.68
Strong β the Client is still largely direct-response dependent (n=40)
Lagged spend effects (T+1 to T+3)
β 0
Weak lag signal β but NetSuite data only 16 mo. Seasonality confounded.
Brand Share vs. Subsequent Brand & NB ROAS β Time Series (40 months)
Green bars = Brand % of paid-search spend. Lines = Brand ROAS and NB ROAS. Watch how high-Brand runs (FebβApr '26 esp.) are followed β or accompanied β by compressed NB ROAS.
Lag Correlation Table β "Does this month predict that future metric?"
Pearson correlation. |r| > 0.5 is a strong signal; |r| between 0.3 and 0.5 is moderate. Negative = inverse relationship.
| Predictor (T) | Outcome (T+K) | T+0 | T+1 | T+2 | T+3 |
|---|
Brand Share (T) vs. NB ROAS (T+2) β Scatter
Each dot is a month. X-axis = Brand % of paid-search spend; Y-axis = NB ROAS two months later. Downward trend is the key finding.
Spend Reduction Events β Follow-Through
Months where total paid spend dropped β₯20% vs. the prior month, with what happened in the 3 months that followed.
| Month | Spend before | Spend after | MoM cut | Next-month NS rev | Next 3-mo avg ROAS | Context |
|---|
π What This Means for the Strategic Question
Hypothesis 1 β CONFIRMED
Brand-heavy months cause subsequent-month weakness in paid-media efficiency, with the effect strengthening over 2-3 months. Specifically, a month with high Brand % share correlates -0.63 with NB ROAS 3 months later (n=37) β the strongest lag signal in the entire 40-month dataset. Pattern held when NS history was extended from 16 β 40 months and after the bidirectional Brand/NB reclassification, meaning it's a real signal β not a sample-size or measurement artifact.
Mechanism (likely): Brand campaigns capture existing intent; they don't generate new awareness. When Brand share is high for sustained periods, the top-of-funnel (NB prospecting) pool shrinks, and two months later both Brand and NB campaigns have fewer fresh searchers to convert.
Q1 2026 is a live example: Feb '26 Brand share 64% Β· Mar '26 66% Β· Apr '26 71%. By Apr '26 NB spend had been cut from $46K (Nov) to $4K (Apr) β a 91% reduction. The data predicts further NB/Brand ROAS compression over May-Jul '26 if Brand-heavy continues.
Brand-heavy months cause subsequent-month weakness in paid-media efficiency, with the effect strengthening over 2-3 months. Specifically, a month with high Brand % share correlates -0.63 with NB ROAS 3 months later (n=37) β the strongest lag signal in the entire 40-month dataset. Pattern held when NS history was extended from 16 β 40 months and after the bidirectional Brand/NB reclassification, meaning it's a real signal β not a sample-size or measurement artifact.
Mechanism (likely): Brand campaigns capture existing intent; they don't generate new awareness. When Brand share is high for sustained periods, the top-of-funnel (NB prospecting) pool shrinks, and two months later both Brand and NB campaigns have fewer fresh searchers to convert.
Q1 2026 is a live example: Feb '26 Brand share 64% Β· Mar '26 66% Β· Apr '26 71%. By Apr '26 NB spend had been cut from $46K (Nov) to $4K (Apr) β a 91% reduction. The data predicts further NB/Brand ROAS compression over May-Jul '26 if Brand-heavy continues.
Hypothesis 2 β NOT CONFIRMED (40-mo dataset)
Spend reductions show weak lag effects. Same-month spend correlates +0.68 with NetSuite revenue, but lagged correlations drop to β0 (T+1: β0.06, T+2: +0.01, T+3: +0.06). With n=40 this is a real null, not a sample-size issue. Possible reads:
(a) The Client's business is primarily direct-response β cuts hurt this month, not future months
(b) Most observed spend cuts are seasonal (post-Jan pullbacks) where the next promo refills naturally
Practical read: Don't be afraid of efficiency-driven spend cuts in the short term, but the Brand-share-lag effect means efficiency cuts concentrated in NB (while holding Brand steady) are worse than proportional cuts across both.
Spend reductions show weak lag effects. Same-month spend correlates +0.68 with NetSuite revenue, but lagged correlations drop to β0 (T+1: β0.06, T+2: +0.01, T+3: +0.06). With n=40 this is a real null, not a sample-size issue. Possible reads:
(a) The Client's business is primarily direct-response β cuts hurt this month, not future months
(b) Most observed spend cuts are seasonal (post-Jan pullbacks) where the next promo refills naturally
Practical read: Don't be afraid of efficiency-driven spend cuts in the short term, but the Brand-share-lag effect means efficiency cuts concentrated in NB (while holding Brand steady) are worse than proportional cuts across both.
What the lag pattern implies
The data suggests proportional spend trimming (across Brand AND NB) avoids the lag-compression effect, while cutting NB while holding Brand reliably triggers the 2β3 month NB ROAS drop.
Q1 '26 did the opposite of proportional β NB was cut aggressively while Brand share rose. The lag analysis predicts a softer MayβJul '26 if this continues.
For the actionable course-correction plan (Brand-share guardrail, NB rebalance target, leading indicators), see the π― Recommendations tab.
The data suggests proportional spend trimming (across Brand AND NB) avoids the lag-compression effect, while cutting NB while holding Brand reliably triggers the 2β3 month NB ROAS drop.
Q1 '26 did the opposite of proportional β NB was cut aggressively while Brand share rose. The lag analysis predicts a softer MayβJul '26 if this continues.
For the actionable course-correction plan (Brand-share guardrail, NB rebalance target, leading indicators), see the π― Recommendations tab.
Caveats on these findings
Β· Correlation β causation. Brand-heavy months may be caused by NB weakness (not the other way around) β e.g. if NB underperforms, the Client might shift budget to Brand, which then makes the Brand-share metric rise while NB ROAS falls.
Β· Seasonality is partly confounded. Q1 months tend to have both Brand-heavy mixes AND naturally weak February/March performance.
Β· Sample size: 40 months of paid-media data AND 40 months of NetSuite revenue. Sufficient for Brand/NB ROAS lag signals and directional reads on revenue lag at T+1 through T+3.
Β· This analysis doesn't factor in promo intensity β a Brand-heavy month that coincides with a big promo (e.g. Feb '26 Presidents Day) might look different to a Brand-heavy month with no promo.
Β· Correlation β causation. Brand-heavy months may be caused by NB weakness (not the other way around) β e.g. if NB underperforms, the Client might shift budget to Brand, which then makes the Brand-share metric rise while NB ROAS falls.
Β· Seasonality is partly confounded. Q1 months tend to have both Brand-heavy mixes AND naturally weak February/March performance.
Β· Sample size: 40 months of paid-media data AND 40 months of NetSuite revenue. Sufficient for Brand/NB ROAS lag signals and directional reads on revenue lag at T+1 through T+3.
Β· This analysis doesn't factor in promo intensity β a Brand-heavy month that coincides with a big promo (e.g. Feb '26 Presidents Day) might look different to a Brand-heavy month with no promo.
Driver #6 β Product Sales Mix
Product/category/size composition of NetSuite revenue itself β this is the bottom-line we care about. Both a driver (what sold) and a downstream result of Drivers 1β5 (what the campaigns pushed).
Evaluate last in the diagnostic walk: if Drivers 1β5 don't explain a NetSuite revenue move, product-level shifts (mattress-order share, attach rate, promo depth, bundling) are the residual story. Source: Adventure the Client Sales CSV (8,104 orders, Jan '25βApr '26).
Evaluate last in the diagnostic walk: if Drivers 1β5 don't explain a NetSuite revenue move, product-level shifts (mattress-order share, attach rate, promo depth, bundling) are the residual story. Source: Adventure the Client Sales CSV (8,104 orders, Jan '25βApr '26).
AOV YoY (Q1 avg)
-27%
Q1 '25 avg $692 β Q1 '26 avg $505
Mattress Order Share YoY
Mixed
Jan '26 down vs Jan '25 (50.6% β 32.1%); Feb '26 down vs Feb '25 (41.3% β 39.0%)
Attach Rate YoY (Feb & Apr)
-36%
Feb '25 2.44 β Feb '26 1.57 Β· Apr '25 2.64 β Apr '26 1.72
Discount Depth Q1 YoY
Down ~3pp
Rules out discounting as cause of AOV drop
AOV β Monthly Trend (40 months)
NetSuite net revenue Γ· unique orders. Now Jan 2023 β Apr 2026.
4-year AOV arc: Stable band $577β$790 from 2023 through early 2025 (3-year avg ~$685). Held through BFCM '25 ($725). Then stepped down sharply: Dec '25 $593 Β· Jan '26 $569 Β· Feb '26 $544 Β· Mar '26 $486 Β· Apr '26 $517. Q1 '26 AOV is ~$170 (~25%) below the 3-year baseline β driven primarily by bundling stopping (attach rate dropped from ~1.5 to exactly 1.0 since Feb '26, meaning every order is single-SKU).
Order Type Mix β Mattress vs Accessory-Only
Stacked count of orders per month by whether they contain a mattress.
Root cause #1: Mattress-containing orders fell from 50.6% of total orders (Jan '25) to 33.5% (Mar '26). More accessory-only orders at lower ticket prices pull AOV down directly.
Accessory Attach Rate
Accessory units per mattress-containing order (protectors, pillows, toppers, sheets, foundations).
YoY signal (not peak-to-trough):
Β· Jan '25 β Jan '26: 1.98 β 2.82 (+42%, up)
Β· Feb '25 β Feb '26: 2.44 β 1.57 (-36%, down)
Β· Mar '25 β Mar '26: 2.26 β 1.99 (-12%)
Β· Apr '25 β Apr '26: 2.64 β 1.72 (-35%, down)
Feb and Apr '26 are materially below their 2025 comparables, but Jan '26 is higher than Jan '25 β so this is not a clean step-change from Jan to Feb. Investigate whether Feb/Apr seasonality alone explains the YoY gaps, or whether a structural change (bundling, checkout, cross-sell) kicked in intermittently.
Β· Jan '25 β Jan '26: 1.98 β 2.82 (+42%, up)
Β· Feb '25 β Feb '26: 2.44 β 1.57 (-36%, down)
Β· Mar '25 β Mar '26: 2.26 β 1.99 (-12%)
Β· Apr '25 β Apr '26: 2.64 β 1.72 (-35%, down)
Feb and Apr '26 are materially below their 2025 comparables, but Jan '26 is higher than Jan '25 β so this is not a clean step-change from Jan to Feb. Investigate whether Feb/Apr seasonality alone explains the YoY gaps, or whether a structural change (bundling, checkout, cross-sell) kicked in intermittently.
Discount Depth β % of Gross
Total discount Γ· NetSuite gross revenue, monthly. Tests the "is deeper discounting causing AOV decline?" hypothesis.
Hypothesis rejected: Discount % averaged ~21% in 2025 and ~17% in 2026 β discounting is actually LOWER in the AOV-decline period. Discounting is not the cause.
NetSuite Category Net Revenue β Monthly
Stacked absolute NetSuite net revenue by product category (mattress / topper / protector / foundation / pillow / sheets).
π Mattress Bundling β Add-On Lift Analysis
Direct measurement of how much value was added to mattress orders via bundled accessories (Mattress + Pillow, Mattress + Topper, Mattress + Mattress Protector + Pillow + Sheets, etc.), from the NetSuite order-combo data.
Jan '25 β Jan '26: bundling was actively happening β add-on value per mattress order ranged $75β$465. Feb '26 onward: the combo breakdown shows zero multi-product orders. Every order in Feb-Apr '26 is single-SKU. Whatever bundling mechanism was in place stopped.
Jan '25 β Jan '26: bundling was actively happening β add-on value per mattress order ranged $75β$465. Feb '26 onward: the combo breakdown shows zero multi-product orders. Every order in Feb-Apr '26 is single-SKU. Whatever bundling mechanism was in place stopped.
Bundled Orders Feb-Apr '26
0
vs 35β305 bundled orders/month through Jan '26
Peak Add-On Value / Bundled Order
$465
Sep '25 Β· 91 bundled orders Β· $42K add-on value
Lowest Pre-Stop Add-On Value
$75
Oct '25 β bundling softening months before Feb '26 stop
Avg Add-On Value / Bundled Order
$125
Consistent $100β168 range across Jan '23 β Jan '26 Β· last recorded Jan '26 Β· zero add-on revenue Feb '26 onward
Bundled Orders Count β Monthly
Count of orders that contained a Mattress + at least one accessory. Volume was variable through 2025; hits 0 Feb '26.
Bundled-Order AOV vs Mattress-Only AOV β Monthly
Top line: mattress-plus-accessories AOV. Bottom line: mattress-only AOV. The gap between them is the bundling lift. After Jan '26 there's no bundled-AOV data point because there are no bundled orders to measure.
The triangulation on AOV decline: three independent data sources agree:
Β· NetSuite attach rate: Feb '26 collapsed to 1.57 accessories per mattress order (vs 2.4-2.9 prior).
Β· NetSuite combo data: zero multi-product orders Feb-Apr '26 (this view).
Β· Google Ads accessory conv value: dropped from $500-900 β $275-312 Q1 '26.
All three are measuring the same operational break from different angles. Something in the Client's bundling/cross-sell mechanic changed in late Jan / early Feb '26 and has not been restored. This is the primary structural lever for AOV and NetSuite AOV-driven revenue recovery.
Β· NetSuite attach rate: Feb '26 collapsed to 1.57 accessories per mattress order (vs 2.4-2.9 prior).
Β· NetSuite combo data: zero multi-product orders Feb-Apr '26 (this view).
Β· Google Ads accessory conv value: dropped from $500-900 β $275-312 Q1 '26.
All three are measuring the same operational break from different angles. Something in the Client's bundling/cross-sell mechanic changed in late Jan / early Feb '26 and has not been restored. This is the primary structural lever for AOV and NetSuite AOV-driven revenue recovery.
π Google Shopping Placement Ad Spend vs. NetSuite Reality
Google Shopping placement performance (PMax & Shopping campaigns, Jan 2023 β Apr 2026) classified as Mattress vs Accessories. Covers only the product-listing-ad (shopping inventory) portion of spend β Search campaigns are excluded.
Lens: is Shopping placement investment driving actual mattress sales on the NetSuite side, or just accessory click-through?
Red-flag metric β Avg conv value per conversion: if mattress placements convert at a sub-mattress-price avg (<$500 per conversion), the ads are attracting cart-adds but not mattress purchases. Same for accessories: normal range is $500-900; below $400 suggests the accessory click wasn't bundled with other items.
Red-flag metric β Avg conv value per conversion: if mattress placements convert at a sub-mattress-price avg (<$500 per conversion), the ads are attracting cart-adds but not mattress purchases. Same for accessories: normal range is $500-900; below $400 suggests the accessory click wasn't bundled with other items.
Mattress Spend Share β Apr '26
70.4%
Peaked at 77% in Apr '24 Β· dropped to 5-20% through H1 '25 Β· climbing again Q1 '26
Mattress ROAS Trend
9x β 4x
Nov '25 9.2x Β· Apr '26 4.0x. Matches historical pattern: ROAS compresses as mattress share >40%.
Accessories Conv $ Collapsed
-60% YoY
2025 H2 avg $650-970 per conv Β· Q1 '26: $275-$312. Sharp break, not gradual.
Mattress Avg / Conv (Q1 '26)
~$1,100
Within DTC mattress price band ($1,200). Mattress ads are driving actual mattress purchases, not cheaper add-ons.
Google Shopping Placement Spend β Mattress vs Accessories (Monthly, 40 mo)
Absolute $ spent on Shopping placements (PMax & Shopping campaigns) by product type. Mattress push emerged mid-2024, accelerated sharply in Q1 '26.
Mattress Share of Shopping Placement Spend β %
% of Shopping placement spend (PMax & Shopping campaigns) targeting mattress SKUs. Historical 5-30%; Q1 '26 spiked to 30-70%.
Shopping Placement ROAS by Product Type
Platform-reported Shopping placement ROAS (has double-count caveat). Watch direction: Mattress ROAS dropping while Mattress spend share rose = diminishing returns.
π₯ Avg Conversion Value per Conversion β "Did they actually buy what they clicked on?"
Shopping placement conversion value Γ· conversion count per product type. Mattress line should sit near $1,200 (DTC mattress price range). Accessories line should sit near $500-900 (typical mattress-buyer accessory basket). Sustained drop below those ranges is the "placement-push vs. actual-purchase mismatch" warning.
Accessory conversion value broke pattern in Q1 '26:
Β· H2 '25 range: $563β$971 per accessory conversion (healthy β implies bundled-with-mattress baskets)
Β· Feb '26: $275 Β· Mar '26: $290 Β· Apr '26: $312 β sudden ~60% drop
Β· Historical 2023βH1 '24 baseline was also low ($120β$500) during high-mattress-share eras β this is a recurring pattern, not a brand-new issue
This aligns precisely with the NetSuite-side attach rate signal (Feb '26 at 1.57 accessories/order vs 2.9+ in prior months). Interpretation: accessory-targeted Google ads are still generating clicks, but those clicks convert to accessory-only baskets instead of adding onto a mattress order. Historically this coincides with periods when mattress ad share is elevated and the accessory budget is squeezed (fewer cross-sell impressions in front of the same buyers).
Mattress conversion quality is intact β Q1 '26 avg mattress conversion value is $1,036-$1,796, squarely within the Client's mattress price band. Mattress ads are driving real mattress purchases, not sub-price orders.
Β· H2 '25 range: $563β$971 per accessory conversion (healthy β implies bundled-with-mattress baskets)
Β· Feb '26: $275 Β· Mar '26: $290 Β· Apr '26: $312 β sudden ~60% drop
Β· Historical 2023βH1 '24 baseline was also low ($120β$500) during high-mattress-share eras β this is a recurring pattern, not a brand-new issue
This aligns precisely with the NetSuite-side attach rate signal (Feb '26 at 1.57 accessories/order vs 2.9+ in prior months). Interpretation: accessory-targeted Google ads are still generating clicks, but those clicks convert to accessory-only baskets instead of adding onto a mattress order. Historically this coincides with periods when mattress ad share is elevated and the accessory budget is squeezed (fewer cross-sell impressions in front of the same buyers).
Mattress conversion quality is intact β Q1 '26 avg mattress conversion value is $1,036-$1,796, squarely within the Client's mattress price band. Mattress ads are driving real mattress purchases, not sub-price orders.
π What Google Ads Product Data Says About the NetSuite AOV Decline
The mattress side is working β but at a cost
Apr '26 mattress share is 70%. Historical precedent: spring '24 hit 77% mattress share (Apr '24), and 2023 averaged ~40%. During the FebβAug '24 heavy-mattress era, mattress ROAS ran 2.5β4.1x; during H2 '25 when mattress share was 20β30%, mattress ROAS was 7β10x. Current Apr '26 @ 4.04x is consistent with the "mattress-heavy = lower-ROAS" pattern.
Mattress conversion quality is intact: avg conv value $1,036β$1,796 across Q1 '26 β people clicking mattress ads are buying mattresses. The mechanic works; the question is scale efficiency.
Caveat: at 70% share with 4x ROAS, the marginal mattress dollar is the most expensive it's been since early 2024. Probably near the ceiling of scalable mattress-intent inventory on Google.
Apr '26 mattress share is 70%. Historical precedent: spring '24 hit 77% mattress share (Apr '24), and 2023 averaged ~40%. During the FebβAug '24 heavy-mattress era, mattress ROAS ran 2.5β4.1x; during H2 '25 when mattress share was 20β30%, mattress ROAS was 7β10x. Current Apr '26 @ 4.04x is consistent with the "mattress-heavy = lower-ROAS" pattern.
Mattress conversion quality is intact: avg conv value $1,036β$1,796 across Q1 '26 β people clicking mattress ads are buying mattresses. The mechanic works; the question is scale efficiency.
Caveat: at 70% share with 4x ROAS, the marginal mattress dollar is the most expensive it's been since early 2024. Probably near the ceiling of scalable mattress-intent inventory on Google.
Accessory side is where the AOV leakage lives
Accessory avg conv value fell from $650β$970 (2025 H2) to $275β$312 (FebβApr '26). Accessory conversion count held steady β what changed is what's in the basket. Accessories are converting to standalone purchases, not add-ons to a mattress order.
This mirrors the NetSuite-side attach rate signal (Feb '26 at 1.57 accessories per mattress order, vs 2.9+ in prior months). Two independent data sources pointing at the same mechanic shift in early Q1 '26.
Historical note: accessory avg conv value also ran low ($120β$500) across 2023 when mattress share was 28β53%. The "accessory conv value compresses when mattress share rises" relationship is recurrent, not new.
Accessory avg conv value fell from $650β$970 (2025 H2) to $275β$312 (FebβApr '26). Accessory conversion count held steady β what changed is what's in the basket. Accessories are converting to standalone purchases, not add-ons to a mattress order.
This mirrors the NetSuite-side attach rate signal (Feb '26 at 1.57 accessories per mattress order, vs 2.9+ in prior months). Two independent data sources pointing at the same mechanic shift in early Q1 '26.
Historical note: accessory avg conv value also ran low ($120β$500) across 2023 when mattress share was 28β53%. The "accessory conv value compresses when mattress share rises" relationship is recurrent, not new.
Strategic tension β where the Client is on the tradeoff curve
Β· Apr '26 mattress ROAS: 4.04x (down from H2 '25 peak of 10.49x in Jun)
Β· Apr '26 accessory ROAS: 5.45x (healthy, comparable to '25 range)
Β· Accessory avg conv value $312 β baskets are smaller than a year ago
Accessories are currently earning higher incremental ROAS than mattress spend, yet mattress share is at a 24-month high. Rebalancing some share back toward accessories β especially around promo anchors (Memorial Day, BFCM) when the attach pattern historically holds β is the lever to pull before pushing more mattress inventory beyond 70%.
Β· Apr '26 mattress ROAS: 4.04x (down from H2 '25 peak of 10.49x in Jun)
Β· Apr '26 accessory ROAS: 5.45x (healthy, comparable to '25 range)
Β· Accessory avg conv value $312 β baskets are smaller than a year ago
Accessories are currently earning higher incremental ROAS than mattress spend, yet mattress share is at a 24-month high. Rebalancing some share back toward accessories β especially around promo anchors (Memorial Day, BFCM) when the attach pattern historically holds β is the lever to pull before pushing more mattress inventory beyond 70%.
What to watch in MayβJul '26
Β· Does accessory avg conv value recover toward the $600β$900 band if mattress share comes down?
Β· Does the FebβAug '24 pattern repeat? Back then, 4β6 months of heavy mattress share (60β77%) was followed by a sharp swing to accessory-heavy in Q4. Watch for a similar forced rebalance.
Β· Memorial Day '26 is the first test: historically Memorial Day pulls mattress orders with strong attach rates. If attach recovers during the promo window, the Q1 '26 attach collapse is promo-timing-related; if it doesn't, the diagnostic shifts to site-side (checkout, cross-sell mechanics).
Β· Does accessory avg conv value recover toward the $600β$900 band if mattress share comes down?
Β· Does the FebβAug '24 pattern repeat? Back then, 4β6 months of heavy mattress share (60β77%) was followed by a sharp swing to accessory-heavy in Q4. Watch for a similar forced rebalance.
Β· Memorial Day '26 is the first test: historically Memorial Day pulls mattress orders with strong attach rates. If attach recovers during the promo window, the Q1 '26 attach collapse is promo-timing-related; if it doesn't, the diagnostic shifts to site-side (checkout, cross-sell mechanics).
π Correlation with the Revenue Anchor β 40-month Pearson r
Computed across Jan '23 β Apr '26 (n=40). |r| β₯ 0.5 strong, 0.3β0.5 moderate, < 0.3 weak.
Mattress order volume IS the NS revenue story
Mattress orders (absolute) β NS revenue: r = +0.99 (T+0) β near-perfect. This isn't surprising: mattresses are the $1,200β1,300 ASP line and drive ~70β80% of dollar volume most months. When mattress order count moves, NS revenue moves.
Β· Mattress orders β NS orders: +0.92 (T+0)
Β· Mattress orders T+3 β NS revenue: +0.33 (weak positive lag β not a reliable signal)
Implication: if you want to predict / steer NS revenue, the mattress-order-count line is the proxy. If mattress orders drop, NS revenue drops. The 40-month view confirms this is the single most reliable relationship in the dataset.
Mattress orders (absolute) β NS revenue: r = +0.99 (T+0) β near-perfect. This isn't surprising: mattresses are the $1,200β1,300 ASP line and drive ~70β80% of dollar volume most months. When mattress order count moves, NS revenue moves.
Β· Mattress orders β NS orders: +0.92 (T+0)
Β· Mattress orders T+3 β NS revenue: +0.33 (weak positive lag β not a reliable signal)
Implication: if you want to predict / steer NS revenue, the mattress-order-count line is the proxy. If mattress orders drop, NS revenue drops. The 40-month view confirms this is the single most reliable relationship in the dataset.
AOV, attach rate, discount % β modest same-month correlations
Β· AOV β NS rev T+0: +0.35 (moderate) β higher-AOV months tend to be revenue months
Β· Attach rate β NS rev T+0: +0.44 (moderate) β high-attach months = higher-AOV = more revenue
Β· Discount % β NS rev T+0: +0.44 (moderate positive) β promo months drive NS revenue this month
Β· Mattress order share % β NS rev T+0: +0.13 (weak) β share of mattress orders doesn't predict NS as well as absolute mattress orders
These are second-order levers vs. mattress order volume itself. AOV and attach are outcomes of the product mix, not primary demand drivers.
Β· AOV β NS rev T+0: +0.35 (moderate) β higher-AOV months tend to be revenue months
Β· Attach rate β NS rev T+0: +0.44 (moderate) β high-attach months = higher-AOV = more revenue
Β· Discount % β NS rev T+0: +0.44 (moderate positive) β promo months drive NS revenue this month
Β· Mattress order share % β NS rev T+0: +0.13 (weak) β share of mattress orders doesn't predict NS as well as absolute mattress orders
These are second-order levers vs. mattress order volume itself. AOV and attach are outcomes of the product mix, not primary demand drivers.
Discount % has a negative T+2 lag β promo pull-forward is real
Discount % β NS rev T+2: r = β0.32 (weak-to-moderate negative) β when discount depth rises this month, NS revenue tends to be slightly lower 2 months later.
Discount % β NS orders T+2: r = β0.38 β even clearer on orders than revenue.
This is the classic promo pull-forward pattern: deep promos move demand from the following 1β2 months forward. The Client's Q4 '25 promo intensity (Nov 27.8% discount) likely contributed to Q1 '26 softness alongside the Q1 spend pullback. Not unique to Q1 '26 β it's a 4-year pattern worth building the promo calendar around.
Discount % β NS rev T+2: r = β0.32 (weak-to-moderate negative) β when discount depth rises this month, NS revenue tends to be slightly lower 2 months later.
Discount % β NS orders T+2: r = β0.38 β even clearer on orders than revenue.
This is the classic promo pull-forward pattern: deep promos move demand from the following 1β2 months forward. The Client's Q4 '25 promo intensity (Nov 27.8% discount) likely contributed to Q1 '26 softness alongside the Q1 spend pullback. Not unique to Q1 '26 β it's a 4-year pattern worth building the promo calendar around.
The Q1 '26 attach-rate drop is the unresolved structural issue
40-month attach rate: range 1.31β1.99 (mean ~1.5) every month 2023βJan '26. Then:
Β· Feb '26: 1.00 (exact 1.0 β no multi-line orders)
Β· Mar '26: 1.00
Β· Apr '26: 1.00
Every order since Feb 1 '26 has been single-SKU. Bundling stopped β this is confirmed and not a measurement artifact. Against the 40-month baseline (attach 1.4β2.0 across all prior promo and non-promo months), this is the sharpest discontinuity in the Client's product-mix history.
Ruled out by the 40-month data: discounting (2023β2025 attach was high under both low-promo and high-promo regimes); seasonality (Feb attach was 1.4+ in both 2024 and 2025). The Feb '26 drop is operational β bundles were pulled from the offer, and that alone accounts for ~$120β150 of the AOV loss vs the 3-year baseline.
40-month attach rate: range 1.31β1.99 (mean ~1.5) every month 2023βJan '26. Then:
Β· Feb '26: 1.00 (exact 1.0 β no multi-line orders)
Β· Mar '26: 1.00
Β· Apr '26: 1.00
Every order since Feb 1 '26 has been single-SKU. Bundling stopped β this is confirmed and not a measurement artifact. Against the 40-month baseline (attach 1.4β2.0 across all prior promo and non-promo months), this is the sharpest discontinuity in the Client's product-mix history.
Ruled out by the 40-month data: discounting (2023β2025 attach was high under both low-promo and high-promo regimes); seasonality (Feb attach was 1.4+ in both 2024 and 2025). The Feb '26 drop is operational β bundles were pulled from the offer, and that alone accounts for ~$120β150 of the AOV loss vs the 3-year baseline.