October 1, 2026
Get named: work the off-site plan
The short version
When an AI engine answers "who should I use for X?", it searches the web, reads a handful of pages and names the businesses those pages mention: review sites, "best X" list articles, Reddit threads, YouTube videos, Wikipedia, press coverage. If your competitors are named and you are not, the usual reason is that they are on those pages and you are not.
So the plan under Improve → Action plan leads with those pages. It has three sections, in this order:
- Get onto the sites AI reads: the sites the engines read when they answered without naming you.
- Win the answers you're already in: the answers where the engines already read or name you.
- Hygiene checklist: collapsed at the bottom. Worth doing once; not a visibility lever.
The page itself is described in the Action Plan Guide. This article is the how-to.
Why off-site mentions matter more than on-site tweaks
The best public evidence points the same way:
- Mentions across the web go with AI visibility. Ahrefs compared about 75,000 brands and found that how often a brand is mentioned across the web tracks its visibility in AI answers (Google's AI Overviews and AI Mode, and ChatGPT) far more closely than its backlinks do: a correlation of roughly 0.66 to 0.71 for mentions against about 0.22 for backlinks, and about 0.74 for mentions on YouTube. (Ahrefs: AI brand visibility correlations, Ahrefs: AI Overview brand correlation)
- Review sites show up in AI answers. SE Ranking found review platforms such as G2, Capterra and Trustpilot in about a third of the Google AI Overviews it studied. (SE Ranking: review platforms in AI Overviews)
- Each engine reads different sites. ChatGPT, Perplexity, Gemini and the others favour different sources, and the sources change from month to month. That is why every task shows which engines read the site, and why the plan is rebuilt from fresh answers.
These studies are correlations, mostly across larger brands, so treat them as direction rather than a guarantee. They agree with what AEOTrack sees in your own answers: the brands named are the brands on the pages that were read.
1. Get onto the sites AI reads
Which sites are listed
A site is listed when, in the last 30 days, the engines read it while answering your questions without naming you, and never read it in an answer that did name you. Your own site, your tracked competitors' own sites, search engines and .gov/.edu sites are left out.
YouTube, Wikipedia and LinkedIn are included. X, Facebook and Instagram are not, because the engines rarely cite them.
How they are ordered
By how often each site was read, weighted by engine (the same engine weights as the score). A read in an answer that named a competitor counts in full; a read in an answer where no competitor was named counts half. A site read only once with no competitor named is left out. The plan shows the top 10 sites.
What each task shows
An off-site task (fictional demo data): the question, the engines and how often each read the site, the competitor named instead, the cited page, the times read, the steps and the status.
- The question the site was read for, and any other questions it was read for.
- Which engines read it, and how often each did.
- The competitor named instead of you.
- The cited page, linked, so you can see exactly what the engines read.
- Times read in the last 30 days.
- How to do it: steps for that kind of site (below).
Statuses
Each task has four statuses:
| Status | Meaning |
|---|---|
| To do | Not started. |
| Contacted | You have pitched, applied or posted, and are waiting. |
| Listed | You are on the page. A Listed task is not reopened for 30 days, to give the engines time to read the page again. |
| Not possible | There is no way onto this site (for example a forum that bans vendors, or a competitor's own page). It stays closed. |
You don't always have to mark Listed yourself: an open task closes itself as Listed once an answer that names you reads that site. Filter the section by status to see what is waiting on a reply.
How to get onto each kind of site
| Kind of site | What to do |
|---|---|
| Review and listing sites (G2, Capterra, Trustpilot, Google Business Profile, app stores, directories) | Claim your profile in the category the cited page uses, complete it, and ask customers for honest reviews. Follow the site's rules: no paid or self-written reviews. |
| List articles ("best X", "X vs Y", "alternatives") | Find the author, and pitch one specific reason to include you that a reader would care about, with facts they can check. |
| Reddit and forums | Follow the community's rules. Reply only where your answer genuinely helps, say that you work for the company, and never ask for upvotes. If vendors aren't allowed to post, mark the task Not possible. |
| YouTube | Offer a review account with no conditions on what is said, and label anything paid as sponsored. Or publish your own clear comparison video. |
| Wikipedia and Wikidata | Notability comes first: the subject needs independent, reliable coverage. Don't edit your own article; suggest sourced facts on the article's Talk page. Keep your Wikidata item factual, with references. |
| Press and trade publications | Pitch a data-led finding a journalist can use, not the product. |
| Founder posts with something worth reading, a complete company page, and useful comments on posts in your field. | |
| Anything else | Find out who runs it, then pitch it, or publish your own page answering the question. If it is another vendor's own page, there is no listing to get: mark it Not possible. |
Work three to five sites a week. A few well-chosen sites beat a long list sent nowhere.
The outreach list (CSV)
On Pro and up, Download outreach list (CSV) exports the section as a spreadsheet with these columns: site, url, source_type, engines, questions, rivals_named, times_read, suggested_action, status. It contains no personal contact details: finding the right person to contact at each site is your step.
2. Win the answers you're already in
The second section works on answers where you are already in the picture:
- Your own pages that AI reads. For each of your pages the engines read, and for which questions, the plan suggests adding statistics, quotes and sources. These suggestions only appear for pages AI already reads; changing a page no engine reads does not change what they answer.
- Short pages are flagged only when the engines cite them, for the same reason.
- Other tasks drawn from your answers, for example questions where the engines read your page but name someone else.
For one question in depth, the opportunity report lays out the answers, the brands named and the pages read.
3. Hygiene checklist
At the bottom, collapsed: schema markup, llms.txt, meta tags and titles, headings, canonical tags, alt text, internal links and robots.txt rules for AI crawlers. These are housekeeping. They are worth doing once, but tests found no measurable effect of schema or llms.txt on whether AI names you:
- Schema markup: a controlled Ahrefs test across 1,885 pages found no significant effect on AI citations. (Ahrefs: schema and AI citations)
- llms.txt: SE Ranking looked at about 300,000 domains and found no clear effect on AI citations. (Search Engine Journal on the SE Ranking study)
Do them once (the GitHub integration can open a pull request for several), then spend your time on section 1.
Check what moved
Checks re-run weekly. After each one:
- See whether tasks you marked Contacted have closed themselves as Listed.
- On the Questions page, watch the questions those sites were read for. The weekly results email and the What changed panel only report moves that clear the ± range, so a change they report is real.
- Pin the questions you are working on (Pin your most important questions) so they are checked daily.
Getting listed and getting named are not instant: the engines have to read the page again, and an article may only be updated once a quarter.