ObsurfableFord can rank in the top 100 for mentions and past 1,000 for citations. Tesco can lead mentions on ChatGPT while trailing on cited domains. Search Engine Land’s reporting on that split is anecdotal at brand scale — Sentic’s State of AI Visibility 2026 (published October 25, 2026) quantifies the pattern across 55 recognizable US brands, five industries, and five AI engines.
The headline for strategists: visibility is concentrated, mentions and citations diverge, and the leader on one engine is often not the leader on another. Blended “AI visibility scores” hide the races you are actually running.
Method in brief
Sentic selected 11 brands per industry in hospitality, senior living, healthcare, B2B SaaS, and ecommerce (55 total). For each brand they pulled October 2026 US data from SE Ranking’s AI Search snapshot (pulled October 24, 2026):
- Brand mentions: tracked prompts whose answer names the brand as an entity.
- Citations: prompts whose answer links to the brand’s website (definition varies slightly by engine).
Share of voice averages each brand’s share of industry mentions and share of industry citations with equal weight — always relative to the 11-brand set, not the whole market.
Engines: Google AI Overviews, Google AI Mode, ChatGPT, Perplexity, and Gemini. Each engine’s prompt corpus differs in size; Sentic warns to compare brands within an engine, not raw counts across engines.
Finding 1: Two brands often hold half — or more — of AI visibility
| Industry | Top-two share of voice (11-brand set) | Leader (SOV) |
|---|---|---|
| Hospitality | 49.9% | Marriott (28.1%) |
| Senior living | 49.8% | A Place for Mom (31.4%) |
| Healthcare | 85.3% | Cleveland Clinic (46.8%) |
| B2B SaaS | 40.2% | Salesforce (26.6%) |
| Ecommerce | 76.6% | Etsy (65.9%) |
Healthcare is extreme: Cleveland Clinic and Mayo Clinic hold 85.3% of share of voice among the 11 systems tracked; the bottom five combined hold 1.7%.
B2B SaaS is the “most open” set Sentic measured — yet the top two still hold 40.2%. That is a different world from Howseen’s SMB baseline where 85.5% of brands were never named on first scan — see Howseen 85.5% Never Named — but the lesson rhymes: AI answers repeat a short list.
Finding 2: Mention-heavy vs citation-heavy brands
Sentic highlights brands that win one metric and lose the other within the same industry set:
| Brand (industry) | Share of mentions | Share of citations | Pattern |
|---|---|---|---|
| Slack (B2B SaaS) | 13.2% | 3.7% | Mention-heavy |
| monday.com (B2B SaaS) | 0.4% | 5.5% | Citation-heavy |
| Chewy (ecommerce) | 4.2% | 15.4% | Citation-heavy |
| Zendesk (B2B SaaS) | 8.5% | 17.7% | Citation-heavy |
| Hilton (hospitality) | 24.7% | 18.8% | Mention-heavy (smaller gap) |
Mention-heavy usually means the model recommends you from entity memory, reviews, or third-party consensus — but sources the answer elsewhere (listicles, retailers, forums). Citation-heavy means your owned pages answer general questions well enough to ground the response even when the prose does not spotlight your brand.
DTC challengers Everlane, Bombas, Glossier, and Allbirds together hold 2.3% of ecommerce mentions and 0.4% of citations in the set — named 3.5–4.8× more often than cited.
That is the enterprise mirror of ChatGPT’s Citation Gap and ghost citations: different remediation for each.
Finding 3: Senior living — operators named, directories cited
| Group | Share of mentions | Share of citations |
|---|---|---|
| Nine operators (e.g. Sunrise, Brookdale) | 75.5% | 24.9% |
| Two directories (A Place for Mom, Caring.com) | 24.5% | 75.1% |
A Place for Mom is named in 3.1% of mentions but earns 59.6% of citations — the most cited site in the set on all five engines.
Operators cannot ignore directory profiles: they are often the first structured entity the model sees. Long-term, Sentic argues for operator-owned pages directories cannot replicate — care levels, staffing, licensing, pricing approach — written as direct answers to family questions.
Note scale: 7,741 senior-living mentions in the 11-brand set vs 542,377 for hospitality hotels — AI footprint varies wildly by vertical.
Finding 4: The hospitality and SaaS leader changes by engine
Hospitality (most mentioned / most cited by engine):
| Engine | Most mentioned | Most cited |
|---|---|---|
| Google AI Overviews | Marriott | Marriott |
| Google AI Mode | Hilton | Marriott |
| ChatGPT | Marriott | Best Western |
| Perplexity | Marriott | Marriott |
| Gemini | Best Western | Best Western |
Marriott leads on most surfaces; Best Western leads Gemini on both measures and is the most cited hotel site in ChatGPT — ahead of Marriott.
B2B SaaS:
| Engine | Most mentioned | Most cited |
|---|---|---|
| Google AI Overviews | Salesforce | Salesforce |
| Google AI Mode | Salesforce | Zendesk |
| ChatGPT | Salesforce | HubSpot |
| Perplexity | Slack | HubSpot |
| Gemini | Slack | Zendesk |
Salesforce leads mentions on Google surfaces and ChatGPT; Slack leads mentions on Perplexity and Gemini. Citations split across Salesforce, Zendesk, and HubSpot depending on engine.
Stable exceptions exist: Etsy leads ecommerce on both measures on all five engines; Cleveland Clinic is the most cited health system on all five.
This is why engine-agnostic GEO scores cannot predict citations and why Pineprompt found only 12.9% of brands named on every major engine.
What to do with the report
1. Split KPIs: mention rate, citation rate, share of voice — per engine
A blended leaderboard can show you “winning” while Slack-style mention-heavy or monday-style citation-heavy gaps persist on the assistant your buyers use.
2. Audit who gets cited for your prompt themes
If directories or comparison sites hold citations, study those pages as templates for extractable structure — not as enemies to spam.
3. Match tactics to your gap type
| Gap | Likely lever |
|---|---|
| Mentions without citations | Entity clarity, reviews, PR, listicle presence |
| Citations without mentions | Answer-first docs, guides, product facts — then entity reinforcement |
| Neither | Authority and third-party evidence before FAQ rewrites — see Page Optimization Only Helps High-Authority Domains |
4. Do not assume Google AI Mode ≡ AI Overviews
Sentic’s tables already show Hilton mentioned most in AI Mode while Marriott leads Overviews. Google’s own guidance says the surfaces can diverge — covered in Google AI Mode vs self-citation shifts.
How Obsurfable fits
Sentic uses a commercial AI Search database for large US brands. Obsurfable publishes prompts, answers, brands mentioned, and citations in an open corpus — Explorer for inspection, the App for matched mention rate, share of voice, and source-to-mention on the questions your buyers ask, per engine.
FAQ
Is share of voice comparable across industries?
No — it is relative to Sentic’s 11-brand slice per industry, not total market demand.
Why equal-weight mentions and citations in SOV?
It forces teams to balance recommendation and owned grounding. If you only care about one, track that metric directly instead of SOV.
Does SE Ranking entity matching miss sub-brands?
Yes — Sentic notes parent brands may not get credit when answers name only a child brand (e.g. Holiday Inn vs IHG).
How does this relate to Search Engine Land’s Ford/Tesco examples?
Same structural split at national scale; Sentic supplies industry tables and engine splits for US categories.
Bottom line
Sentic’s October 2026 snapshot shows concentrated AI visibility, systematic mention/citation divergence, and engine-specific leaders — even among famous brands. Measure each race separately, on the engines your buyers use, then fix the bottleneck you actually have: entity recommendation, owned citations, or both.