# AEO Diligence Questions The tests that separate an AEO company from a dashboard feature or an agency with software. ## Feature, company or agency - **Tracker only?** Then it is a feature. Ahrefs bundles it, and Adobe, HubSpot and Sitecore bought it. - **Who does the work?** "We own the outcome" often means done-for-you content. Ask for gross margin and services share of revenue. - **Published price vs real contract.** A large gap between list tiers and enterprise deals usually means services. - **Average contract.** Divide ARR by customer count. It shows if an "enterprise" story is real, or if small accounts carry the revenue. - **Logo overlap.** Check if pipeline logos also sit on the cap table. ## Is the data a moat - **Scale against peers.** Ahrefs reports 239M prompts a month. Compare any "citation records" claim with that. - **Compounding type.** Learning per client gives switching costs. Only learning across clients gives a network effect. - **Sample size.** Tuning many weights on a few dozen articles, with noisy AI answers, risks overfitting. - **Coverage per client.** Ask how many prompts a flagship client tracks. Total record counts can hide thin coverage. - **First-party data.** Search Console, GA4 and CDN logs link AI visibility to real traffic. This is harder to copy than public citation data. ## Is the result real - **Control group.** Before/after shows change, not cause. Ask for untouched control prompts. - **Durability.** Recency is a gatekeeper factor, so fresh pages can win fast and fade. Ask for 90-day retention. See [[What Drives AI Citations]]. - **Category depth.** Winning a thin category proves less than winning a contested one. - **Prompt source.** No platform shares real user prompts. "Real questions" are usually modelled. Ask how they are built. ## Is the data collection safe - **API or interface.** API answers can differ from what users see. Peec and Profound both drive the consumer interface, so this is table stakes, not a moat.^[https://docs.peec.ai/] - **Legal exposure.** Interface capture can mean bypassing bot detection. Reddit's suit against SerpApi and Perplexity mostly survived dismissal in July 2026.^[https://www.law360.com/articles/2508228] Regulated clients will ask about this. - **Location.** Answers change by geography. Check that tests run from the buyer's market. - **Dependence.** Check exposure to one research or search provider. See [[Agent-Ready Web Stack]]. ## Is the product still needed in two years - **Platform risk.** A model provider could sell its own AEO tools. The best defence is neutrality across all models, since no provider will measure its rivals. - **Ads.** ChatGPT Ads may take budget from organic AEO. - **Agents.** Does the roadmap reach agent-readiness and commerce, or only content? See [[Agentic Commerce Protocols]]. - **Exits.** Disclosed exits so far: XFunnel ~$30M, Scrunch ~$225M (reported). Entry price must work at that range. ---- tags: #boat #ai #investing links: [[Agent Engine Optimisation MOC]] [[AEO Company Landscape]] [[AI era Defensibility]]