Our AEO Build Log: Making JYNLAB Show Up in AI Search (Documented)

We use JYNLAB as the test case for our own AEO method, in public. Baseline: 0 of 4 AI engines cited us, and all four had a different wrong identity. The documented, step-by-step log of what we did, and the checklist you can steal.

We sell AEO, so the most honest proof we can offer is to do it on ourselves in the open. This is the running log: the baseline, every change, and what we are measuring. It is also the checklist we run for clients. For the method behind it, see our AEO method; for the concepts, the complete guide to AEO.

The baseline: what we found

Before changing anything, we ran our buyer prompts through ChatGPT, Perplexity, Gemini, and Claude. The result was a clean zero, and worse, an entity mess:

• 0 of 4 engines named JYNLAB for AEO prompts like "who does AEO for B2B SaaS". They named competitors instead.
• 4 engines, 4 different wrong identities: one read an old "app market intelligence" identity off our site, one confused us with a similarly-named medical-imaging company, one hallucinated a CRM product, one knew only an outdated "growth agency" description with no external verification.

The lesson was immediate: this was not a content problem first. It was an entityproblem. The engines could not even agree on who we are, because our own site and our third-party footprint were sending old, conflicting signals.

What we did, in order

Step 1: Fix the entity on-site

We aligned every owned signal to one definition, "JYNLAB is an AEO agency for B2B companies". That meant rewriting the homepage title and description (they still said the old "growth marketing automations"), the About page, and adding an Organization schema in the raw HTML so crawlers resolve the entity reliably. We also 301-redirected the www host to the apex so the entity is not split across two sites, and noindexed a legacy product page that was broadcasting the old identity.

Step 2: Build the topic cluster

We built one pillar (the complete guide to AEO) and linked spokes around it: AEO vs SEO, how LLMs choose citations, how to get cited in ChatGPT, a commercial "best AEO agency" page, and a vertical page for our lead niche. All cross-linked hub-and-spoke so the engines read us as the source on the topic, not a scatter of posts.

Step 3: Make it extractable, with schema

Every page leads with a direct answer, uses clear headings, lists, and tables, and ends with an FAQ. We added Article, FAQPage, and ItemList structured data, and a Service schema on the offer page. Schema does not make you citable on its own, but it helps engines parse and trust what is already clear.

Step 4: Set up measurement

We defined a fixed prompt set across category, commercial, and vertical intents, and a log to record, per engine, whether we are cited, who else is, and which sources got pulled. This is the share-of-model tracker we re-run on a cadence. You cannot improve what you do not measure.

Step 5: Off-site consensus (in progress)

The baseline showed competitors were cited through third-party sources, not their own sites. So the current phase is building consensus: consistent profiles on the directories and review sites engines trust, genuine contribution in the communities our buyers read, and inclusion in comparison roundups. This is the slower, compounding lever, and it is where the real lift comes from.

The checklist (steal this)

If you are doing this yourself, this is the order that compounds fastest:

1. Run your buyer prompts through the engines and record the baseline. Note who is cited.
2. Make every owned signal say one clear thing about who you are (title, meta, About, Organization schema, one canonical host).
3. Fix conflicting old signals (noindex legacy pages, redirect duplicate hosts).
4. Build one pillar plus linked spokes that answer your buyers' exact questions, answer-first.
5. Add Article, FAQPage, and Organization schema. Keep it crawlable and server-rendered.
6. Make your entity consistent off-site: LinkedIn, Crunchbase, Wikidata, G2, same definition.
7. Earn third-party mentions where buyers and engines both look.
8. Re-run the prompt set every two weeks and let the gaps steer the next move.

What we are measuring next

The entity fixes are live; the next crawl is what lets the engines re-resolve who we are. We re-run the same four-engine prompt set on a fixed cadence and will update this log with the result, whether it moved or not. An honest "not yet" is more useful than a dressed-up number. That is the same standard we hold our client reporting to.

Frequently asked questions