How AI engines choose who to recommend, and what you can do about it
When a buyer asks ChatGPT, Claude, Perplexity, Gemini or Grok to recommend a product, the engine mostly repeats what the independent pages it reads for that question say. A well-built website makes you easy to describe once you are found. Being named on the pages the engines read is what puts you on the shortlist. AEO Analyzers measures both, and keeps a dated record of what you do about them.
AEO Analyzers is built for marketing teams and agencies that need to show their work: every answer stored, being found by name kept separate from being recommended, the engine panel named on every result, and a dated record of what changed between measurements.
It comes down to three steps: Measure. Act. Record.
1. Measure: where do you stand today?
You get two numbers, kept separate:
- Ready to be cited. Your AEO Score: whether engines can reach your site, read it and describe you correctly. This part is under your control.
- Recommended by engines. Your Citation Sweep: how often engines name you when a buyer asks the category question, with your name nowhere in it. Every answer is stored, and every result names the engines it came from.
You also get Source coverage: the pages the engines actually read for your category questions, and whether each one names you yet.
2. Act: who does what?
Every action in your report is tagged with the team that owns it.
- Site (your web team): crawl access, clear facts and structured data. This is usually done once.
- Content (your writers): pages that answer the questions buyers ask, with the answer first.
- Outreach (your PR or partnerships lead): the pages engines read that do not name you yet, such as directory listings, industry roundups, reviews from real customers, and findings you publish that others can cite.
For most established companies the site work is already in good shape, and outreach is where the shortlist is decided.
3. Record: what changed, and when?
Mark each action done, with its date. Your next sweep, on the same engines and the same questions, shows its figures beside those dates. Month after month, that becomes a record you can hand to a board or a client: what you did, when you did it, and what was measured afterward.
Our own example
On October 13th, 2026, four engines found aeoanalyzers.com by name in 30 of 40 answers and recommended it in 0 of 200 category answers. Our site was ready, and we were not yet on the pages the engines read for our category. We publish our own record month by month at https://aeoanalyzers.com/evidence, including the months a number goes down.
What we promise, and what we do not
No method can guarantee that an engine will recommend you, and we do not claim one. We do promise that every number we report comes with its date, its engines, its denominator and the answers behind it, so you and anyone you share it with can check it.
Questions
Why does a good website not get me recommended?
A good website makes you easy to describe once an engine finds you. For a category question, engines mostly draw on independent pages such as roundups, directories and reviews. If those pages do not name you, the engine has nothing to repeat.
What should I do first?
Run an AEO Score to confirm your site is ready, then a Citation Sweep to see which pages the engines read for your category. Start with the pages cited most often that do not name you yet.
How will I know if it is working?
Measure again on the same engines and the same questions. Your Record shows each action's date beside the figures from every sweep, so a change is visible against what you did, not guessed at.