Mentions or Citations: Which Lever Should You Fix First?

Fix mentions first if AI engines already cite your site as a source but never mention your brand in the answer. Fix citations first if the reverse is true: your brand gets mentioned, but no citation points to your site. Most sites have both gaps, and each one needs a different fix.

What's the difference between a mention and a citation?

A mention is your brand name appearing in the text of an AI answer. A citation is a link to your domain as the source behind that answer. They sound like one signal from the outside. In practice they're two separate systems that happen to share the same interface.

For the full mechanical breakdown, see What Is a Ghost Citation? and Mentions vs. Citations.

Why don't mentions and citations move together?

A citation happens when the AI decides your page backs up a specific claim, then links back to it as the source. A mention happens through a separate decision entirely: whether your brand name belongs in the text of the answer itself.

Two decisions, made at two different points in how the answer gets built, and a site can win one without winning the other. When a site wins the citation and loses the mention, that gap has a name: a ghost citation.

How do you find out which lever is broken for you?

You can't tell from a rankings dashboard, because rankings don't measure either signal. Checking it by hand takes 15 to 20 real buyer questions and roughly that many minutes each. CitedScore's AI Visibility Report runs the same classification at scale: 20 prompts, four platforms, 80 observations, aggregated so one weird answer doesn't skew the read.

The output is a grid showing which prompts cited you, which mentioned you, and where the two split apart.

Which lever should you fix first?

Fix whichever gap is larger in your own results. Citations and mentions respond to different repairs, and fixing the wrong one first wastes a cycle. Count your split, then work the bigger number down. If your citations outnumber your mentions, you have a recognition problem.

The engines trust your content enough to cite it, but they don't associate your name with the answer. That usually traces back to thin entity signals off-site, how consistently your brand gets referenced across the sources these engines pull from when they're deciding who to mention. Fix that, and the citations you already have start converting into mentions.

If your mentions outnumber your citations, you have a structure problem instead. The engines already know your brand, but they aren't citing your site as the source. That usually points at how your content is organized for extraction: headings, answer placement, chunkable sections. Fix that, and the mentions you already have start earning the citation.

What does fixing each lever involve?

Fixing a recognition problem means building the entity signals an AI relies on to recognize your brand outside your own site. That includes consistent business descriptions across directories and profiles, a presence on hubs like Crunchbase or your industry's association pages, and consistent naming, so a model can tell three mentions across the web refer to the same business.

Fixing a structure problem means changing how your content is built for extraction. That includes front-loaded answers near the top of each page, question-based headings that match how buyers phrase things when they type into a chat box, and specific, isolated claims that a model can lift and attribute back to your page.

This distinction matters past the diagnosis. A mention puts your name directly in front of the buyer, inside the text they're already reading. A citation puts a link in front of them, one they may or may not click. Citation without mention does the analytical work and gets none of the credit for it.

A software company might show up cited on ten technical comparison prompts and mentioned on only two of them, a clear recognition gap. A regional service business might get mentioned confidently on local-intent prompts but never cited as the source behind a specific claim, a clear structure gap. Same underlying issue, opposite fix, and the split tells you which one you're looking at before you spend a quarter guessing.