The AEO Discovery Audit: A 14-Day Process for Mapping Buyer Query Intent

Why Discovery Comes Before Publishing

The temptation in AEO is to start publishing on day one. A blog editor, a list of category keywords, and a writer can produce 30 articles in three weeks. The output looks like progress. The citation report at day 90 looks like failure.

The reason is almost always topic mismatch. The 30 articles answered questions buyers do not ask AI, in language buyers do not use, framed around problems buyers do not feel acutely enough to type into a chatbot. The pages exist. The queries do not.

The discovery audit closes that gap. Two weeks of structured query gathering and ranking, before the first article ships, produces a roadmap where every article is tied to a real query with measurable demand.

The 14-Day Structure

The audit fits in two weeks for one analyst working full-time, or three weeks for a part-time effort. The structure is the same regardless of pace.

Each phase has a discrete output. Skipping any one of them creates a blind spot. Buyer interviews catch the language buyers use. Sales calls catch the actual decision queries. Competitor reverse-engineering catches the queries already being answered (and which the brand is losing). AI probing catches the queries the brand wants to win.

Phase One: Buyer Interview Synthesis

The audit starts with four to six interviews with current or recent buyers. The interview format is not a customer research interview, it is a query archaeology interview.

The interviewer asks the buyer to recall the specific phrases they typed or said to AI when researching the category. Not the topic in their head, the literal words. Most buyers can recall three to six specific queries from a recent purchase. Aggregated across six interviews, the output is 25 to 40 real queries in real buyer language.

The pattern that always emerges is that buyer queries are shorter, more direct, and more solution-shaped than marketers expect. "Best CRM for healthcare consulting" beats "customer relationship management software comparison guide" every time. The audit captures the former and rejects the latter.

Phase Two: Sales Call Transcript Mining

The second source is sales calls. Twenty to forty recent calls (won and lost, across stages) yield another 40 to 80 queries.

The mining technique is specific. The analyst searches transcripts for verb-led phrases (how, what, why, can, should, when) followed by the brand's category language. Each hit is a candidate query. After deduplication and rephrasing into search-natural form, the sales call set typically overlaps 30 to 50 percent with the buyer interview set, validating the high-confidence queries.

The non-overlapping queries are where sales calls add unique value. Late-stage, objection-handling, and evaluation queries (security, compliance, integration) appear in sales calls but rarely in early buyer interviews. These queries are gold for AEO because the corresponding articles tend to convert downstream, even if they earn fewer citations than top-of-funnel queries.

Phase Three: Competitor Citation Reverse-Engineering

The third source is competitor citations. The analyst runs the top three competitors through a citation tracking tool, extracts the queries where each competitor is cited, and adds them to the candidate set.

This phase often surfaces queries the buyer and sales call sources missed. Competitors typically have stronger top-of-funnel content that AI surfaces, including queries that buyers ask before they realize the category exists ("how do I track AI mentions of my brand" leading to AEO category awareness, for example).

The reverse-engineering also tags each query with current competitive density. A query already cited by three competitors is more expensive to enter than a query cited by one. The tag feeds into priority scoring at phase six.

Phase Four: AI Model Probing

The fourth source is direct probing of AI models. The analyst takes the candidate set and runs each query through the four major AI models (ChatGPT, Claude, Gemini, DeepSeek), records which models surface citations at all, who they cite, and what the answer shape looks like.

This phase reveals model-specific patterns. Some queries return list answers (good for table-structured content), some return narrative answers (good for prose explainers), some return comparison answers (good for side-by-side tables). The structure of the future article changes based on what shape the model returns for the query.

The probing also reveals which queries are uncitable today. If the AI model returns "I cannot answer that without more context", the query is too vague or too narrow to drive AEO investment. These queries get cut from the roadmap.

Phase Five: Deduplication and Clustering

By day eight, the analyst has a candidate set of 200 to 400 queries from four sources. Deduplication cuts it to 120 to 200 unique queries. Clustering groups them into 8 to 15 topic clusters.

A cluster is a query group that shares a topic, a buyer intent, and likely a citation source. "How does AEO work", "what is answer engine optimization", and "how do I get my brand cited by ChatGPT" cluster together. A single foundational article often answers the cluster, with three to five supporting articles deepening specific angles.

The clustering is the moment the publishing plan starts to write itself. Each cluster needs a foundational article, a measurement article, a case-study article, a how-to article, and an objection-handling article. Five articles per cluster times 10 to 15 clusters is the first 50 to 75 articles of the program.

Phase Six: Priority Scoring

Day eleven assigns a priority score to each cluster. The scoring rubric is simple and consistent.

The output is a ranked list. The top three clusters get the first 25 to 30 articles. Clusters in the bottom third get deferred to year two.

Phase Seven: Roadmap and Publishing Plan

The final two days convert the ranked clusters into a publishing plan. The plan specifies the first 60 articles by slug, cluster, query, persona, and target publish week.

The plan is the deliverable the editorial team works from. It is also the artifact that locks alignment with the brand, compliance (where applicable), and leadership before publishing starts. Most disagreements about AEO direction happen in week two of the engagement, when the plan is reviewed, not in month six when the citations are coming back wrong.

OnlyAEO ships the 14-day discovery audit as a fixed deliverable. The roadmap that comes out of it is the publishing plan for the first 90 days of the engagement, with quarterly updates based on citation performance.