Citation share

What is citation share?

Where attribution rate measures one source against all queries, citation share measures one source against other sources for the same queries. The basic formula: (target source citation instances) ÷ (total citation instances across all sources for those queries) × 100. Example: if 100 queries about "GEO" produce 250 citation instances total across all sources, and your domain is cited 35 times, your citation share is 14%.

Citation share is a practitioner-coined GEO metric, the AI-search analog of traditional share of voice (measuring relative presence rather than absolute volume), but the two are not mechanically equivalent. Traditional SEO share of voice is built on keyword rank or impression positions; brand-analytics share of voice is built on counted media mentions. Citation share is built on counted citation events in AI responses. The frame is similar; the units, data sources, and tooling are different. Vendor and academic literature do not define this exact operationalization for AI search; the term and the per-engine / per-query-cluster operationalization below are crystallized from GEO measurement practice (Profound, Otterly, and similar tools each implement slightly different internal definitions, see caveats below).

Before reporting any number, lock three axes:

• URL vs domain vs brand: example.com/page-a (URL), example.com (domain), and "Example Corp" (brand entity including all owned domains and profiles like YouTube channels) yield different numbers.
• Deduplication: same domain mentioned multiple times in one AI answer can be counted raw (each citation counts) or deduped per-answer (one per answer). Choose one and document it.
• Per-engine vs cross-engine aggregation: see the FAQ below for the three common options.

Status in 2026

Critical KPI for competitive GEO analysis¹². Used by enterprise SEO teams to argue for budget ("we own 22% citation share on AI Overview for the [topic cluster]") and by indie practitioners to identify uncontested topics where citation share is concentrated in just one or two players. The latter signal is especially valuable: a topic with 80% citation share owned by a single competitor is hard to break; one fragmented across many sources is enterable.

How to apply

Citation share answers "how big is my slice of the AI citation pie for this topic?" Three operational moves:

• Define the topic cluster before measuring: citation share is meaningless without a fixed query set. Pick 20 queries that scope your target topic cluster, freeze the set for ~8 weeks, then measure each engine's citation distribution across all sources. Note engine coverage: Perplexity and Google AI Overview surface citations consistently; ChatGPT / Claude / Gemini surface citations only when web search or grounding is enabled, so per-engine numbers from these surfaces need measurement-coverage caveats.
• Tabulate competitor presence, not just your own: spreadsheet columns: query / source URL / source domain / brand entity / engine / week. Roll up to whichever granularity matches your reporting purpose (URL for optimization decisions, domain for category comparisons, brand for executive reporting). Practitioners frequently report that AI-engine competitor sets surprise them vs. classic SEO competitor sets (different sources surface for the same query intent).
• Hunt for fragmented topic clusters (practitioner heuristic, calibrate by engine and query-set size): if no single source holds more than ~15% citation share for your cluster, it looks enterable. If one domain holds 60%+ share, you're competing against an entrenched canonical; pick a different cluster or wait for the canonical to age out. These thresholds are heuristics, not industry benchmarks.
• Calibrate share targets to the surface's citation slot count: citation share competition is zero-sum within each surface's slot pool, and slot counts vary by an order of magnitude across the citation-surfaces cluster. DuckDuckGo Search Assist surfaces only 1-2 linked sources per answer (the tightest slot in the cluster, per the DuckDuckGo AI citation entry); Perplexity surfaces ~10-20+ sources (the widest); ChatGPT search, Microsoft Copilot, and Google AI Overview sit between with source-card panels typically displaying 5-15. A 22% citation share on DuckDuckGo Search Assist (1 of 4-5 queries) reflects much tighter competition than 22% on Perplexity (4 of 20 sources). When benchmarking competitor presence, normalize against the surface's slot pool rather than treating "22% share" as a comparable number across surfaces. See the cross-surface quick reference in AI citation metrics for slot counts per surface.

What to skip: paid citation-share dashboards (Profound, Otterly.AI, Peec AI, AthenaHQ, Brand Radar) until you have 30+ priority queries to track. Manual spreadsheets work fine at lower volume.

How it relates to other concepts

• Competitive frame of attribution rate: attribution rate measures absolute frequency across queries, citation share measures relative slice within the same query set.
• Conceptual analog (not mechanical equivalent) of traditional SEO and brand-analytics share of voice KPIs. The frame is similar; the underlying unit (citation event vs ranking position vs media mention) differs.
• Distinct from brand mentions in AI answers, which counts your absolute presence including unlinked mentions; citation share counts the relative slice of linked citations only.
• Combined with citation match rate to assess both presence (share) and quality (linked) of citations.
• Inversely correlated with topic saturation: fragmented topics offer entry; concentrated topics signal a defended position.
• Denominator alignment: in the Citation vs Mention vs Link taxonomy, citation share counts citation cells (linked + unlinked citation instances) normalized against total citation instances across all sources for the prompt set, not mentions. Brand mention share is a separate metric on the mention column.