The Goal
The Client (a personal-injury law firm focused on motor-vehicle accidents) had a meaningful asset sitting in their MCC: a former lead-gen partner's Google Ads account — a calculator-driven lead-aggregator that had spent $41.7M across 1.17M unique search queries over three years. The data had been accessible for some time. We recently consolidated it to mine three years of PI-auto search-demand intelligence — paid for and tested by someone else — for what it actually contained.
The two accounts run different businesses in the same demand pool:
· The aggregator was selling leads (B2B). Their hook was a "Pain & Suffering Calculator," their conversion was a form fill, and they competed in the cheaper consumer-tool auction. 3-yr avg CPC: $8.76. 3-yr avg CPA: $99.83. · The Firm is selling legal representation (B2C). Their hook is contingency-fee representation, their conversion is a qualified lead (must be in an accident + injured + not at fault + seeking an attorney), and they compete in the premium legal-services auction. L12M avg CPC: $77.04. L12M avg CPA: $787.
The job: extract the intelligence and use it to restructure the Firm's own $2.45M/year account. The brief had three parts:
· Where in the demand pool is the Firm under-invested relative to what the aggregator proved out? · Where is the legal-services CPC premium structurally smallest — i.e., where are the keyword-arbitrage opportunities? · What does an account architecture look like that concentrates spend on those arbitrage clusters without abandoning the attorney-keyword auction the Firm currently dominates?
How We Went About It
Three layers stacked: source the data, define the arbitrage lens, build the restructure.
Data sourced and joined across both accounts:
· Aggregator: 6,234 keywords across 326 ad groups, 1.17M unique search terms triggered, 4.78M clicks, 418K conversions tracked, 1,029 ads analyzed for creative patterns, 2,592 logged account changes · Firm: 608 keywords across the L12M window (May 2025 – April 2026), $2.45M spend, full search-term + ad + change-history exports · Joined: 87 overlapping keyword+match-type combinations where both accounts bid, used for head-to-head CPC and CVR comparison
The slice itself — question vs. non-question intent — wasn't arbitrary. The Firm had started testing question-based keywords in their existing MVA Question campaign and was seeing strong performance: $24.66 avg CPC, $336 CPA over three months on $16,820 of spend, well below the $77 / $787 blended account baseline. That validated the intent cluster locally. The aggregator dataset answered the next question — how much of this is actually scalable? — by sizing the demand pool and CPC profile across the same intent at three years of investment.
The arbitrage lens was the analytical move. CPC gaps between the two accounts aren't aspirational ("the Firm should run cheaper"); they're diagnostic. Where the gap is smallest, the legal-services premium is structurally smallest — which means the Firm has the best relative auction position. Where the gap is largest (up to 13.9x on exact-match attorney terms), the legal-services premium is brutal and the Firm is paying it regardless of how it bids.
Reporting was generated in Python (build_account_restructure.py, build_keyword_comparison_v3.py) so every chart and table traces back to the underlying CSV exports. Two PDFs and one restructure workbook were the original deliverables; this self-contained HTML consolidates the analysis into a single navigable file.
The restructure itself was built on a signal-based keyword framework — every query scored by how many qualifying signals appear (Lawyer Need / Not at Fault / Injured / Claim Value). Single-signal queries land in an informational campaign; two- and three-signal queries land in a consideration campaign with ad groups by signal pair. Three-signal queries are the qualified-lead candidates and get the most aggressive bids.
Key Insights
(1) Question-based intent is where the legal-services CPC premium is structurally smallest. The CPC gap by intent tier: 2.7x on exact-match question keywords (smallest in the dataset) vs 13.9x on exact-match attorney terms (largest). Same demand pool, two different auctions — the Firm's best arbitrage position is on question intent, not attorney terms.
(2) The Firm was over-concentrated on the worst-arbitrage segment. 97.7% of the Firm's $2.45M was running on attorney/lawyer terms (565 keywords, $79 avg CPC, $789 CPA). Question-based keywords carried 2.3% of spend (43 keywords, $35 avg CPC, $710 CPA). The aggregator deployed $2.10M on question intent over three years — proving the demand exists at scale, even after translating through the Firm's stricter qualified-lead bar.
(3) There's $1.67M of validated demand the Firm has zero coverage on. 1,041 question-intent keywords run in the aggregator account that don't appear in the Firm's account at all. At the aggregator's $9.18 avg CPC and the smallest gap multiple (~3x for legal services), expect ~$27 CPC on those clusters — well below the Firm's blended $77 and far below the $139 exact-match attorney-term price.
(4) On a handful of hyper-specific lower-funnel terms, the CPC premium inverts. Keywords like "i got rear ended and my neck hurts" actually cost the Firm less than they cost the aggregator (0.5–0.9x gap). The aggregator's broad-funnel buyers don't want those queries; the Firm's qualified-lead buyers do. Worth concentrating into a dedicated ad group with matched landing.
Recommendation: shift ~$66K/month from attorney-term auctions to question-based intent. Three new campaigns — Question Multi-Signal ($38K, primary market), Question Informational ($20K, transformed from existing MVA campaign to preserve conversion history), Multi-Market Question ($18K, consolidated across secondary markets). Hold the total spend envelope; change the exposure profile. Switch question campaigns from Max Conversions to Max Conversion Value once each hits 15+ signed cases/month — the threshold at which value-bidding has enough signal to outperform conversion-bidding on contribution margin.