What Is Citation Share?

How Citation Share Works

A generative engine answers a question by retrieving a handful of documents and synthesizing them into a single response, usually attaching citations to the sources it leaned on. Citation Share counts those attributions. For a given topic, you tally every source cited across a set of answers, then ask how many of them are yours. The result is a percentage: your slice of the total citations that engine handed out.

That framing borrows directly from share of voice, the traditional metric for your slice of visibility in a channel. The difference is what gets counted. Share of voice counts impressions, clicks, or mentions; citation share counts cited sources inside AI answers. Because generative engine optimization targets inclusion in a synthesized answer rather than a ranked link, citation share is the metric that tells you whether that work is paying off. A related measure, selection rate, asks how often you are cited when your page is eligible; citation share instead asks what fraction of the whole citation pool you own.

Two properties make citation share behave differently from a search ranking. First, it is defined against a prompt set — the questions you choose to measure. Change the questions and the number changes, because the pool of competing sources changes. Second, it is engine-specific. The same query returns a different set of grounding sources on ChatGPT, Perplexity, and Google AI Overviews, so a domain that dominates one engine can be invisible on another. Reported together, citation share across engines is a core component of AI visibility.

Formula

Citation Share is a straight ratio:

Citation Share (%) = (Your cited sources for a topic ÷ Total cited sources for that topic) × 100

The pieces that make it meaningful:

• Numerator — the count of AI-answer citations pointing to your domain across the measured prompts.
• Denominator — every citation surfaced for that topic, across every source, not just your competitors.
• Scope — the prompt set. The number is only comparable against the same list of questions over time.
• Engine — the platform. Share is computed and reported separately for each engine, because sources do not transfer between them.

A citation is an attributed source — a link or explicit source reference — not a passing brand mention. Being named in the answer text while a rival’s page is cited as the source counts for the rival, a distinction that separates citation share from an AI brand mention.

Example of Citation Share

A documented worked example comes from a 2026 citation study by SolCrys, which recorded 36,268 citations across five AI engines — ChatGPT, Perplexity, Google AI Overviews, Gemini, and Claude — over 30 days (May 12 to June 11, 2026), using 44 buyer-comparison prompts in the Answer Engine Optimization category. Because every citation was logged against a fixed prompt set and a fixed engine list, the dataset is a clean citation-share measurement rather than a loose tally.

Applying the formula to that pool, reddit.com earned 2,882 citations, so its citation share is 2,882 ÷ 36,268 ≈ 7.9% — the largest single slice in the study. wikipedia.org followed with 1,489 citations (≈ 4.1%), and techradar.com with 1,133 citations (≈ 3.1%). No domain broke 8%, which is the headline lesson: even the most-cited source in a category holds a small absolute share, because the citation pool is spread across thousands of long-tail pages. In the same dataset, long-tail domains together accounted for 16,368 citations — roughly 45% of the total — leaving the named leaders competing for what remained.

The scoping matters as much as the ranking. These shares describe one category (AEO buyer-comparison queries) over one 30-day window across five specific engines. Rerun the study on a different topic, a different set of prompts, or a single engine and the leaderboard reshuffles — the same reason a domain’s citation share cannot be quoted as one universal figure. What the example demonstrates is the mechanic: pick a bounded set of questions, count every attributed source, and divide. That is citation share, and it is the only way to say concretely whether generative engines are citing you or citing someone else for the questions you care about.

Citation Share vs Share of Voice

The two are the same idea built for different eras. Share of voice measures presence in a channel of links and impressions; citation share measures being quoted inside a synthesized answer. As more queries resolve to an AI response instead of a list of links, citation share becomes the version of the metric that reflects where attention — and attribution — actually lands.

Frequently Asked Questions

The Bottom Line

Citation Share tells you what fraction of the sources an AI engine reaches for, on a specific topic, belong to you. It is share of voice rebuilt for a world where the answer is synthesized rather than ranked: define the questions, count only the cited sources, and read the result per platform. Chase share on the prompts your buyers actually ask, not a single flattering average.