A mention is when an AI engine names your brand somewhere in its answer text. A citation is when the engine points to a specific source — usually a URL — as the evidence behind a claim. Most brands track the first and ignore the second, which is a mistake: they are produced by different mechanisms and, in our benchmark, they told very different stories about who actually controls AI-mediated trust in the insurance category.
Two Different Signals, Two Different Optimization Plays
Carriers dominate mention volume by sheer scale: 89 of the 137 tracked brands are carriers, accounting for 82.9% of all organic mentions (3,588 of 4,328). Publishers are a tiny slice of the brand universe by comparison — just 5 of 137 brands and 4.9% of mentions (211 of 4,328) — but they punch far above that weight in citations, with 156 citations traced back to those 211 mentions, a 73.9% citation rate. Publishers that build content specifically for extraction — comparison tables, sourced rankings, dated methodology — get cited in a very high share of their own mentions: MoneyGeek in 100.0%, Bankrate in 97.0%, NerdWallet in 53.2%. Carriers, whose content is written for compliance and human browsing rather than extraction, get cited far less consistently per mention.
But raw citation frequency isn't the whole story. Our methodology also strips out self-citation — cases where a source is simply linking back to itself — to measure independent authority. Once that's removed, the picture flips: NerdWallet's citations are only 7.3% independently backed, and MoneyGeek's only 27.9%. Carriers, cited less often overall, are backed independently far more consistently: Amica at 43.5%, Nationwide at 43.3%, Travelers at 41.2%, Erie Insurance at 39.5%, Chubb at 37.1%, USAA at 35.6%.
Why the Two Metrics Diverge
Mentions are largely a function of entity recognition and category presence — how often and how clearly a brand name surfaces across an engine's training data or live retrieval. A well-known incumbent gets mentioned by sheer weight of prior exposure, independent of whether its own content is well-structured for citation.
Citations are a function of retrieval and corroboration — which specific documents an engine's retrieval layer pulls, ranks, and is willing to attribute a claim to. That favors domains with clean structured data, specific numeric claims, and citable formatting. But independent citation — corroboration from a source that isn't citing itself — is the signal that actually functions as third-party trust, and our data shows publishers dominate raw citation volume while carriers dominate independent backing.
| Entity | Type | Mentions | Gets cited | Independently backed |
|---|---|---|---|---|
| MoneyGeek | Publisher | 61 | 100.0% | 27.9% |
| Bankrate | Publisher | 33 | 97.0% | 18.2% |
| NerdWallet | Publisher | 109 | 53.2% | 7.3% |
| Amica | Carrier | 124 | 44.4% | 43.5% |
| Nationwide | Carrier | 252 | — | 43.3% |
| Travelers | Carrier | 170 | — | 41.2% |
| Erie Insurance | Carrier | 38 | 39.5% | 39.5% |
| Chubb | Carrier | 140 | — | 37.1% |
| USAA | Carrier | 270 | 38.1% | 35.6% |
Source: Brainpan.AI Insurance AI Visibility Index™ — Q2 2026 benchmark, 137 brands, Finding 04 (Citation Authority).
What Each Metric Actually Predicts
Mention volume predicts awareness — whether a brand shows up in an answer at all. It does not reliably predict whether the engine recommended that brand: in our top-10 leaderboard, State Farm leads mention volume by a wide margin (457 mentions) but converts only 40.9% of those into active recommendation language, while USAA, with 187 fewer mentions, converts 46.3% — the highest recommendation rate in the top 10.
Citations, and specifically independently-backed citations, correlate more strongly with the kind of third-party corroboration buyers actually trust. Practically: track mentions to know if you exist in the conversation. Track citations — and who's actually vouching for you independently — to know if you're the reason the engine trusts what it said.
Frequently Asked Questions
Is a citation always more valuable than a mention?
Not always, but for high-intent, comparison-style queries a citation carries more weight because it signals the engine treated that source as evidence rather than background context. Mentions still matter for awareness in educational or exploratory queries where no single source is cited.
Can a brand get citations without getting mentioned?
Yes. In our Q2 2026 Insurance AIVI benchmark, Perplexity was the engine most likely to cite a brand's domain as a source without naming the brand in the visible answer text. That is a real but distinct visibility dimension — the content contributed to the answer without the brand receiving credit in the reader-facing response.
Why do publishers get cited more often than carriers, even when carriers are named more?
Publishers like NerdWallet, Bankrate, and MoneyGeek build content specifically structured for citation, so they get cited in a very high share of their own mentions. But much of that citation is self-referential: NerdWallet is cited in 53.2% of its mentions, yet only 7.3% of those citations remain independently backed once self-citation is removed. Carriers are cited less often per mention, but when they are, the citation is far more likely to be genuinely third-party — Amica at 43.5%, Nationwide at 43.3%.
How is a citation different from a backlink?
A backlink is a hyperlink placed by another site, indexed and weighted by traditional search engines. A citation is an AI engine's own decision, at answer-generation time, to attribute a claim to a specific source. A domain can have strong backlink authority and still earn few AI citations if its content isn't structured for extraction.
Should enterprise brands optimize for mentions, citations, or both?
Both, but they require different work. Mentions are improved by broad entity presence and category-relevant PR; citations are improved by publishing structured, sourced, extractable content. And a citation alone isn't the finish line — our data shows the more valuable signal is an independently-backed citation, not just any citation.
See where your brand stands
The AI Visibility Audit benchmarks your brand's citations, mentions, and independent-authority rate against named competitors across ChatGPT, Gemini, Perplexity, Copilot, and Claude.
Request AI Visibility AuditPrefer to browse first? Download a sample audit (PDF) →
Kevin Walsh
Kevin Walsh is the founder of Brainpan.AI, where he builds AI visibility infrastructure, GEO/AEO strategy, schema systems, and citation optimization programs for brands that need to be retrieved, cited, and trusted by AI answer engines.