Your page is in the footnotes. Someone else is in the answer. This is a specific and fixable situation, and the fix is not more content, which is what everyone tries first.
What is actually happening
The model does two separate things when it answers "who should I hire for X".
It retrieves passages relevant to the question, from anywhere. Your page qualifies, which is why it is in the citations.
Then it decides which businesses to name. For that it wants evidence that a business is real, established and worth pointing someone at. A page saying you are excellent is not evidence, because every page says that.
So a well-written page gets you quoted and does nothing for the second decision.
How to confirm this is your situation
Ten prompts, logged out, two columns: named, and cited.
If your domain appears in citations noticeably more often than your name appears in answers, this is you. We measured our own site at 48 citations and 8 mentions across 608 answers, a ratio of six to one.
If neither column has much in it, this is not your situation and the advice below does not apply. Your pages are not being retrieved and that is a different and more tractable problem.
The four causes, in order of how often we see them
1. You are not in anyone else's content. The sources deciding who gets named are round-ups, comparisons, industry lists and community threads. If none of them mentions you, the model has your page as a source and no third-party evidence you are a real option.
2. Your citation sources are incomplete. The registers, directories and review platforms your category runs on are how a model checks legitimacy. Incomplete or inconsistent entries read as uncertainty about which business is real.
3. A competitor owns the comparison. If the only page answering "alternatives to X" or "best X for Y" is a competitor's, they are writing your description. This is most acute in B2B software and increasingly common everywhere.
4. Your content answers the topic and not the choice. A page explaining what a service is gets retrieved for the explanation. A page helping someone choose gets you named in the choosing. These are different pages and most businesses only have the first.
What actually closes it
Get into other people's content. Round-ups, comparison pieces, industry lists. This is slow, it has a long feedback loop, and it is the thing that moves the number. Three to six months is realistic.
Publish something worth citing. Original data is the most reliable route into someone else's article, because it gives them a reason to mention you that is about their piece rather than about you. It is also the only route you fully control.
Be present where your buyers talk. In the category we measure, Reddit is cited 225 times across 608 answers, ahead of every agency website. YouTube at 145 and LinkedIn at 134 also beat every vendor site. Those are not link plays, they are the places the model is reading.
Finish the unglamorous listings. Especially the category-specific ones rather than the general ones everybody claims.
What does not close it
More blog posts. If you are already being read, being read more does not change the second decision. This is the single most common wasted year in this field.
Better keyword targeting. There is no ranking to improve.
A site redesign. The model is reading text.
The uncomfortable framing
Being in this position means your content is working. It is genuinely good enough that a model chose it to build an answer from. What is missing is that nobody else has vouched for you, and vouching is not something you can publish.
That is why this fix takes longer and feels less productive than writing. It is also why it is the thing worth starting first, because the clock on it is longer than on anything else you could do.
The order to work in
Because this is the slowest gap to close, the sequence matters more than in any other situation.
First, confirm it is actually your situation. Run ten questions logged out and record cited and named separately. If you are cited in several and named in none, this is your problem. If you are cited in none either, this article is not about you and the fix is faster.
Second, claim and correct before you pitch. Being mentioned somewhere new is worth much less if the register entry a model checks against still has your old business name on it. This takes two weeks and it makes everything after it work better.
Third, create one reason to be cited. Original data, a genuinely useful tool, a resource that answers something nobody has answered properly. One is enough. This is what converts outreach from asking for a favor into offering something.
Fourth, go where the citations already are. Not everywhere. The specific sources your category's answers already draw from, which you can read straight off the citations in your own test.
Fifth, wait. Three to six months. The work compounds and the reporting will look flat for a while, which is the part that makes people stop.
What closing this gap looks like
“Customers were asking ChatGPT and Gemini for product recommendations in our exact category, and we weren't even in the conversation. Within three weeks of launching Omni Eclipse's AEO strategy, we were the most-cited brand in the space - ahead of long-established brands with far more online presence.
PROCERT is a building certification firm that had grown entirely on referrals. Within a couple of weeks of the work going live, someone who had never met them inquired directly through the website. Fur Magic went from absent to the most-cited brand in their category in three weeks, ahead of long-established competitors with far more online presence.
Both published with the starting number, the end number and the period: PROCERT and Fur Magic.
A worked example
A specialist recruitment firm, cited on nine of twenty tracked questions and named on one.
What that felt like. Failure. They had published consistently for two years, the content was genuinely good, and the answer kept naming three larger competitors. The natural response, and the one they had been following, was to publish more.
What the citations actually showed. Their own guide to salary benchmarking in their niche was being cited as the source for an answer that then recommended someone else. Their content was doing unpaid work for competitors, which is the precise shape of this problem.
What the competitors had that they did not. Presence in four industry round-ups, two association member lists, and a long-running community thread. Not better content. More third-party mentions.
What they did. Stopped publishing for a quarter. Claimed two association listings. Pitched three round-ups and landed two. Answered questions in one community, without links, for twenty minutes a week.
Where it landed. Named on five of twenty by month five. The citation figure barely changed, because it was already high, and that was the point.
The hardest part. The quarter with no publishing, during which every visible metric looked flat and the actual work was happening somewhere they could not see.
See which questions name you
A free AI visibility audit on your own buying questions, across all five answer engines.
Book your audit
Ashur Homa
Built and scaled a digital brand to $100M+ in sales with zero ad spend. Has helped businesses generate millions through AI go-to-market strategy. Leads growth at Omni Eclipse.
Connect on LinkedIn
