The Law Firm That Could Not Prove AEO Was Working
Marlowe & Reed is a 22-attorney regional law firm with practice areas in family law, estate planning, and small-business litigation. Six months ago, marketing added FAQ pages, attorney bios, and schema to the site. Google Business Profile calls are up 18%, but the managing partner cannot see AEO’s contribution: GA4 shows a small amount of traffic from chatgpt.com and perplexity.ai, much of the rest lands as direct/none, and intake staff sometimes hear “I found you through ChatGPT.” The firm tracks keyword rankings for 150 terms but has no AI prompt set, no citation tracking, and no way to connect an AI-assisted visit to a signed matter. The partner wants a one-number ROI before renewing the budget. Marketing has four weeks to propose a measurement plan. The firm’s CRM is intake-based, and call tracking is inconsistent across offices.
1. Design a measurement framework with leading and lagging indicators for this firm.
Read the worked answer
Lead with indicators the firm can observe now, and treat revenue as a lagging, imperfectly attributed outcome. Leading indicators: coverage and correctness on a defined prompt set (for example, 30 questions across the three practice areas and five offices), citation share on those prompts, branded search volume, AI-referred sessions where referrers are visible, and assisted conversions from AI-referred visits. Also track intake mentions: add a required “How did you hear about us?” field with an AI-assistant option.
Lagging indicators: qualified consultations, signed matters, and revenue by practice area and office. These matter, but they should not be attributed solely to AEO without a comparison. Use the CRM’s intake source field and call tracking, and fix the inconsistency across offices before drawing conclusions.
The framework should distinguish “presence” from “impact.” Presence is whether Marlowe & Reed appears and is correct in AI answers. Impact is whether that exposure correlates with branded search, qualified consultations, and signed matters. Report both, with confidence levels and known gaps.
2. How should the team handle dark traffic and self-reported attribution without overclaiming?
Read the worked answer
Dark traffic is expected in AI-assisted journeys: a person may see an answer, search the firm’s name later, call directly, or arrive through a messaging app. The team should not force every direct/none session into an “AEO” bucket. Instead, triangulate three sources: referral traffic where the platform passes a referrer, self-reported attribution at intake, and branded search lift in the firm’s service areas.
Implement a short intake script: “How did you first hear about us?” with options for Google search, AI assistant, referral, directory, social, and other. Train intake staff to record it consistently. Add UTM parameters only where the firm controls the link; do not invent them for AI platforms. For call tracking, standardize the numbers and routing across offices so a call can be tied to a location and campaign.
Then present attribution as a range, not a precise number. For example: “Visible AI referrals are small, but 14% of new intakes in the last quarter self-reported an AI assistant, and branded search rose in two service areas.” That is honest and actionable. Claiming exact ROI from dark traffic would undermine trust in the measurement plan.
3. What should marketing report at 90 days, and what should it explicitly not claim?
Read the worked answer
At 90 days, report leading indicators and data-quality improvements: prompt-set coverage, percentage of correct and cited answers, branded search trend, visible AI referrals, self-reported AI mentions at intake, and the number of offices with consistent call tracking. Include a sample of actual answers and citations so the partner can see what “visibility” looks like. Report lagging indicators as directional, with denominators and caveats.
Do not claim a precise return on investment, causal lift in signed matters, or that every direct visit came from AI. Do not compare this quarter’s revenue to last quarter’s without accounting for seasonality, intake capacity, and practice-area mix. Do not present rankings as proof of AI visibility.
Instead, propose a 6–12 month evaluation with a pre-registered prompt set, a consistent intake field, and a comparison group such as offices or practice areas with different AEO investment. State what would change the conclusion: if citation share rises but qualified consultations do not, the content may be visible but not persuasive. That framing gives the partner a defensible decision rather than a false one-number answer.