The market leader is missing from the answer.
Three buying questions. Three brands built for those exact segments. None of them in the answer.
Last week I ran a small experiment: three questions a B2B software buyer would actually type into an AI assistant. One engine, one run, one date. Not the full audit — a quick check of what the answer looks like.
What we asked
| Buying question | Who ChatGPT named | Who was missing |
|---|---|---|
| Best CRM for a 20-person B2B SaaS sales team | HubSpot, Pipedrive, Attio, Salesforce, Zoho, Close | A CRM built specifically for small businesses |
| Best project management tool for product teams | Linear, Jira, Productboard, ClickUp, Asana, Notion, monday | A tool built around connected product work |
| Best MDR provider for small and mid-sized businesses | Huntress (#1), Arctic Wolf, Sophos, eSentire, Red Canary, Blackpoint, CrowdStrike | No obvious gap — the leader defended its place |
Two of the three segments were missing a brand that was built for them. The third showed what it looks like when a brand owns the query: it is named first, and the alternatives are framed around it.
Why this happens
AI does not rank authority. It assembles an answer from the pages that fit the question. In a 150-query study of US B2B software searches, vendor websites earned 44.1% of citations — often a competitor’s “best X” roundup. YouTube took 20.5%, publishers 13.3%, and Reddit/forums 10.4%. Review sites were only 5%.
And the number that should stop you: only 28.7% of the pages AI cited also ranked in Google’s top 10. A #1 ranking is a weak proxy for being in the answer.
The three pages that do the work
When a brand is absent, the cause is usually structural, not editorial. The page shaped like the question does not exist — so the answer comes from someone else’s page that is.
- “Best [category]” — a real roundup with a comparison table, not a sales page. This is the page AI cites when the buyer asks for the shortlist.
- “Alternatives to [you]” — honest, with use cases and limits. It captures the buyer who already knows your name but is comparing.
- “[You] vs [competitor]” — a table, not an essay. Comparison queries are where deals are won or lost before the demo.
Add one 3-minute video and two genuine community answers in the threads AI already cites. YouTube and Reddit together account for about a third of the source mix in the study.
A 30-day sequence
| Week | Ship | Why |
|---|---|---|
| 1 | Best [category] page | Owns the shortlist query |
| 2 | Alternatives page | Captures comparison-stage buyers |
| 3 | [You] vs [competitor] | Defends the deal before the demo |
| 4 | One video + two community answers | Adds the source types AI cites |
Before you build anything
Run the 12-question scorecard. It takes two minutes and shows which of these pages you already have — no email needed to see the score.
Take the scorecardIf you want the full picture
The AI Visibility Audit runs 288 answers across ChatGPT, Perplexity, Gemini and Google AI Overviews, compares you with five competitors, diagnoses the sources that won, and gives you 10 prioritized fixes. Five business days. €390 for the first five clients.
Method note: the three-question check is one engine (ChatGPT), one run, on 4 November 2026. It is a directional signal, not a measurement. The 150-query figures come from AI Ranked / Zenodo, “AI Overview Citations for B2B Software Queries (US, Oct 2026, Wave 1)”. No client results are claimed. The audit measures public AI answers and public pages; it does not guarantee rankings, citations, traffic or revenue.