Measurement

Citations: the pages behind the answer

A mention tells you an assistant named you. A citation tells you which page convinced it. For most teams, the citation list is the most actionable thing an AI visibility tool produces.

When an AI assistant searches the web before answering, it reads a handful of pages and often lists some of them as sources. Those listed pages are citations. They are the closest thing AI search has to a ranking you can act on, because each one is a real page that someone can update, pitch or publish.

From question to cited page

  1. The question“Best accounting app for a bakery?”
  2. Hidden searchesThe assistant runs its own sub-queries.
  3. Pages readIt reads several results from those searches.
  4. Pages citedSome of those are shown as sources in the answer.

Ahrefs calls those hidden sub-queries fan-out queries. Its Found in measure records every page that was cited or just retrieved in the background, while citations count only the pages shown as sources. That distinction is worth knowing: a page can shape an answer without being credited.

Mentions and citations are different wins

Mention

Your brand’s name appears in the answer text.

  • Shows the assistant knows you
  • Can come from training data alone
  • Hard to act on directly

Citation

A page is listed as a source for the answer.

  • Shows which page influenced it
  • Usually needs live search
  • Points at something you can change

The two don’t always travel together. You can be mentioned because a review site describes you, while your own site is never cited. Or your page can be cited for a fact, while the answer recommends a rival. Tools report both: Otterly, for instance, separates brand coverage (answers mentioning you) from domain coverage (answers citing your site).

Who is being cited?

A long citation list is only useful once it is sorted. Peec AI’s pricing page describes classifying sources as competitor, editorial, reference or user-generated content. Whatever your tool calls them, these groups lead to different actions:

Your own pagesKeep facts current and answer the questions buyers actually ask.
Rivals’ pagesSee what question they answer that you don’t.
Editorial and reviewsComparison articles and reviews: a PR and outreach target.
Reference sitesDirectories and encyclopaedic pages: check your listing is accurate.
Forums and communitiesReal user discussion. Listen first; don’t astroturf.
Everything elseTrack it. A new source appearing often explains a sudden change.
TakeawayEach week, find the pages cited for your most important questions where a rival appears and you don’t. That short list is usually the best to-do list an AI visibility tool can give you.

Reading citation numbers carefully

  • Not every answer has sources. Answers written from training data alone often list none, so citation rates depend heavily on platform and search mode.
  • Platforms change what they share. Scrunch’s guide notes, for example, that one platform stopped contributing citation data while still contributing mentions.
  • Counts and coverage differ. “Ten citations” might be one page cited ten times or ten pages once each. Look for both.
  • Being cited isn’t being chosen. A cited page can support an answer that recommends someone else. Read the answer too.

Our guide to reading the evidence behind the numbers pairs well with this. If your own pages are never cited, start with the basics: can AI search crawlers reach them at all? Our guide to AI crawlers and robots.txt covers that check.

Sources

  1. Ahrefs: AI visibility metrics viewed 11 Nov 2026
  2. Ahrefs: How to view fan-out queries generated by AI viewed 11 Nov 2026
  3. OtterlyAI: Brand report KPI definitions viewed 11 Nov 2026
  4. Scrunch: Metrics guide viewed 11 Nov 2026
  5. Peec AI pricing (source classification) viewed 11 Nov 2026

Directory figures come from Hyperfind’s profiles of 34 tools, not the whole market. Prices follow each provider’s published offer on the date shown in its profile.

Tools in this story

Price your workload ↗Compare tools ↗

Keep reading

Measurement

What an AI visibility tool actually measures

Strip away the dashboards and every tool is counting the same few things in an AI answer. Here is the anatomy, and the questions to ask about each number.

ChartWorked example
For agencies

An agency’s AI visibility week, day by day

Tracking is the easy part. The value is in what happens between Monday’s data and Friday’s client note. A practical weekly routine, with a suggested time budget for each step.

Worked example
Websites

AI crawlers and your robots.txt, without the myths

Blocking every “AI bot” can quietly remove you from AI search answers. Blocking only the training crawlers is the opt-out most businesses actually want. The crawlers, what each one does, and a decision tree for your robots.txt.

Decision treeWorked example