Which AI Surfaces Actually Cite Us on Commercial-Intent Prompts, and Why Is One Blended AI Visibility Score the Wrong Instrument for That Decision?

Which AI surfaces actually cite us on commercial-intent prompts, and why is one blended "AI visibility" score the wrong instrument for that decision?

Your board deck has one "AI visibility" number. Your buyers ask ChatGPT. Those are different instruments. Foundation x AirOps measured citation share by surface. The blended score fails the measurement job.

Disclosure (read first): This report was produced by Foundation Marketing Services in collaboration with AirOps; Foundation has a commercial relationship with AirOps. Numbers below are theirs; we score the buyer decision, not the vendor.

What the study publishes

Scope (Foundation x AirOps, confidence: study page). Commercial-intent B2B software prompts from Jun 22 to Jul 24, 2026 across six AI surfaces: ChatGPT, Claude, Gemini, Perplexity, Google AI Overview, and Google AI Mode. AirOps logged roughly 380,000 AI answers. Foundation classified about three million citation events across 81 B2B software categories. Source: Foundation lab

Citation definition (required). A citation event is one domain appearing as a source in one AI answer. It measures AI-answer visibility, not traffic and not conversion.

Two retrieval systems. Google AI Overview and Google AI Mode build from Google's live search index. ChatGPT, Claude, Gemini, and Perplexity run as standalone assistants with their own retrieval and training data. Product domains lead on every surface. The split shows up in reviews, communities, social, and media.

Social gap (~27x). About 7% of citations on Google's two AI surfaces go to social domains (YouTube, LinkedIn). On ChatGPT, Gemini, and Claude, social's share is 0.3% or lower. The report calls that roughly a 27x gap between AI Overview and ChatGPT. (confidence: study page)

Community fork. ChatGPT community citation share is 9.26%, driven almost entirely by Reddit. Claude community sits at 2.67%. (confidence: study page)

Gemini is vendor-heavy. Gemini sends 71.63% of its citations to product domains. No single third-party domain appears in even 14% of Gemini's answers. (confidence: study page)

G2 on five surfaces. G2 appears in about a quarter to a third (~25-33%) of citing answers on the other five surfaces. Gemini is the outlier. (confidence: study page)

Category swings the mix. Across six representative categories in the study, reviews' citation share swings 3.3x (7.64% managed databases to 25.07% recruiting/ATS). Community share swings 2.5x (5.5% SEO platforms to 13.89% EDR/XDR). (confidence: study page)

Comparison / listicle URLs (soft-tag). Some secondary coverage cites "84 of 100 most-cited URLs" as comparison/listicle pages. That figure was not verified on the public lab page we fetched for this decoder. Treat it as soft until confirmed in the full report PDF. Do not put it in a board slide as proven.

AirOps cross-vertical 68.4% (separate study). AirOps' own May-June 2026 cross-vertical analysis puts B2B software brand-and-product citation share at 68.4%. That uses a different taxonomy. It is not interchangeable with the six-surface Foundation x AirOps mix. Tag it AirOps proprietary when you quote it. (confidence: study page, labeled separate)

Prior Foundation x AirOps 57M work is a different method. Do not mix those method counts with the ~380K / ~3M figures above.

Proof Gap: blended score vs surface-specific citation

The Proof Gap here is instrument failure. Teams buy or report a single AI visibility score. The study shows surfaces diverge on social, community, and third-party share. A healthy Google AI Overview number can hide a ChatGPT hole. A ChatGPT win can hide a Gemini miss. Citation rate and citation share must be tracked per surface. Anything else is spend without proof on the surface buyers use.

Decision rule for Director+ teams (RevOps, enablement, revenue systems)

1. Kill the single blended AI visibility score as a board metric. Track citation rate and citation share per surface across the six named surfaces.
2. Match the dataset to the surface. Google AI Overview / AI Mode need a live-index view. ChatGPT needs community weight. Gemini needs a product-domain view.
3. Treat G2 like owned media on five surfaces. ~25-33% of citing answers outside Gemini. Operate the profile.
4. Fund the contested share with category context. Reviews and community swing hard by category (3.3x / 2.5x in the study's six-category cut). Copying another category's playbook funds the wrong channel.
5. Prompt set = how buyers ask. Recurring commercial-intent prompts per surface. Turn off one-time blended audits.
6. Buy visibility tools only if they break out by surface. If the vendor cannot show ChatGPT vs Google AI vs Gemini separately, you are buying the Proof Gap.
7. Do not convert citation share into traffic or pipeline in the deck. Visibility only. Prove traffic and conversion with your own analytics.

Eight-Seat Read (lite): one decision each

1. CMO / demand: Kill the single blended AI visibility dashboard this quarter?
2. Content / AEO: Fund product and comparison pages before the next editorial sprint?
3. RevOps: Which of the six surfaces do buyers in our category actually use?
4. Product marketing: Treat G2 as owned media?
5. Community / social: Fund Reddit or YouTube only if category and surface data say so?
6. CRO: Refuse board slides that show one AI visibility number?

Turn off: one-score OKRs; category-agnostic playbooks; one-time audits instead of recurring prompt sets.

Vendor claim score

/5: "Track with continuous prompt tooling / one platform sees all six." Surface-split measurement is the useful claim. An AirOps (or any vendor) sales pitch that collapses six surfaces into one score is not. Do not pick their chart as proof of purchase. Demand per-surface citation rate and citation share on your category prompts.

Why we might be wrong

1. The study window is about five weeks (Jun 22-Jul 24 2026). Mixes can move.
2. Google AI Overview does not trigger for every query, so its answer volume differs structurally from the standalone assistants.
3. Commercial co-authors (Foundation Marketing Services and AirOps). Read numbers as published methodology; score the buyer decision yourself.
4. Citation share is not pipeline. A surface win can still fail conversion.

Sources

1. Foundation x AirOps, The State of B2B AI Discovery 2026: Foundation lab. Study window Jun 22-Jul 24 2026; ~380K AI answers; ~3M citation events; 81 B2B software categories; six surfaces (ChatGPT, Claude, Gemini, Perplexity, Google AI Overview, Google AI Mode).
2. Disclosure: This report was produced by Foundation Marketing Services in collaboration with AirOps; Foundation has a commercial relationship with AirOps. A citation measures AI-answer visibility, not traffic or conversion.

Soft CTA

If you want the next AEO decoder and the weekly receipts before you buy another visibility score, subscribe