Citation building
Getting cited by AI search engines
A growing share of searches now end in an answer rather than a list of links. ChatGPT, Perplexity, Google AI Overviews and AI Mode all write their answer from a handful of sources they cite. If your brand is not among those sources, you are not in the answer, no matter where you rank.
What the data actually says
0.66 to 0.71
Correlation between branded web mentions and AI visibility, across ChatGPT, AI Mode and AI Overviews
Ahrefs, 75,000 brands, December 2025
0.194
Correlation for number of pages on your site. Publishing more pages is close to unrelated to being cited
Ahrefs, same study
38%
Share of AI Overview citations that come from a page ranking in the top 10 for that query, down from 76% a year earlier
Ahrefs, 863,000 keywords and 4 million AI Overview URLs
19.6%
Reddit’s share of mentions among cited domains, second only to YouTube
Ahrefs Brand Radar, June 2026
Why rankings stopped being the whole story
A year ago, three quarters of the pages cited in an AI Overview also ranked in the top ten for that query. That number is now closer to a third. The rest is split almost evenly between pages ranking 11 to 100 and pages ranking nowhere at all.
That cuts both ways. Page one no longer guarantees you a place in the answer, and being off page one no longer rules you out. What the models appear to reward instead is whether they can find your brand discussed, in context, across sources they already trust.
This is why the correlation data is so lopsided. Branded mentions across the web correlate strongly with AI visibility. Raw backlink counts correlate weakly. The number of pages you publish barely correlates at all. Volume of content is not the lever. Being talked about is.
What a citation is made of
Your brand named in someone else’s writing
Models build an idea of what your brand is from how other people describe it. A sentence on a real editorial page that names you and says what you do is worth more than the same words on your own site, because it is independent.
Repetition across independent domains
One mention is an anecdote. The same association repeated across unrelated sites becomes a pattern, and patterns are what retrieval picks up. Spread matters more than any single placement.
Topical proximity
A mention on a site that already covers your subject connects you to that subject. A mention on an unrelated site connects you to nothing in particular, even if the domain is strong.
Recency
These systems lean toward recent material when a query has a current answer. Citations decay. A page that was cited last quarter is not guaranteed to be cited this quarter, which makes a steady drip of new mentions worth more than one large push.
Where RankJet fits
We are a marketplace for editorial placements on real, indexed sites. That is the mentions half of the work described above. Here is how we would use it.
Pick sites that already cover your subject
Filter the marketplace by niche and language rather than by Domain Rating alone. Topical fit is doing more work here than authority. A mid authority site that writes about your category every week is a better citation source than a strong site that has never mentioned it.
Write so the mention is quotable
Name the brand plainly, say what it does, and put a specific claim near it. Answers get assembled from sentences that stand on their own. Vague copy that never names you gives a model nothing to lift.
Spread placements across domains
Several placements across unrelated sites beat several placements on one site. You are trying to create agreement between independent sources, and that only works if the sources are independent.
Keep publishing
Because citations decay, treat this as a monthly cadence rather than a campaign with an end date. A handful of fresh placements every month holds visibility better than a burst that then goes stale.
What we cannot promise
Nobody can guarantee an AI citation. There is no submission form for ChatGPT and no ranking dial to turn. What we sell is a placement on a real site, published and indexed. Whether a model then cites it is outside anyone’s control, including ours, and any vendor telling you otherwise is guessing.
The correlation figures above are correlations, not promises. They tell you where to spend effort, not what you will get.
Editorial placements are also only part of the picture. The same research puts YouTube first among cited sources and Reddit second. We do not sell either, and we would rather say so than imply our marketplace covers the whole problem. If your category lives on Reddit, spend time on Reddit as well as here.
Questions we get
Is this different from link building?
It overlaps but the target is different. Link building optimises for a crawler following a href. Citation building optimises for a model deciding which sources to quote, and that decision leans more on the words around your brand name than on the link itself. In practice a good editorial placement does both jobs at once.
Do I need to rank first?
No. Only about a third of AI Overview citations now come from pages ranking in the top ten for the query, so ranking and being cited have come apart. Rankings still matter for clicks from classic results. They are no longer the entry ticket to the answer.
How many placements does it take?
We will not invent a number. The honest answer is that repetition across independent domains is what moves it, so think in terms of a steady monthly cadence rather than a threshold you cross once. Start small, track your brand in the tools that monitor AI answers, and scale what shows up.
Can I measure it?
Partly. Tools now track how often a brand appears in AI answers, and that is the number to watch. It is noisier than rank tracking and it moves for reasons you did not cause, so read the trend over months rather than day to day.
Does the nofollow attribute matter here?
Less than it does for classic link equity. A model reading a page sees your brand named in context regardless of the link attribute. That is one reason mentions correlate better than backlink counts.
Start with sites that already cover your subject
Filter by niche and language, check the editorial guidelines each publisher set, and build from there.