Among 2,661 buyer questions with at least 20 recommendations in Parse’s October 19, 2025–April 25, 2026 ChatGPT and Google AI Overviews sample, the median question had 10 recommended brands and a leading brand with under 25% of recommendation weight. The wider study covered 612,952 recommendations; the concentration finding applies to the eligible question subset.
How many brands compete for the average AI buyer question?
Parse breaks every AI answer down to the topic behind it ("high-yield savings account," "visual sales pipeline management," "truck accident legal representation") and records which brands the model recommends for that topic. Across the 2,661 topics with enough volume to measure reliably, the median had 10 brands competing, with a typical range of 7 to 14. The leading brand for a question held a median of 24.8% of its recommendations. More than four in five of these questions (86.1%) had at least six brands circling, and just 12.2% were locked, meaning one brand took half or more of the recommendations. Half (50.1%) were wide open: no brand above 25%. The headline most teams get wrong is that being recommended at all feels like progress, when in most categories it only puts you in a crowd.
- The median AI buyer question has 10 brands competing for it, and the leading brand holds a median of just 24.8% of the recommendations.
- 86.1% of well-tracked buyer questions have 6+ brands competing; only 12.2% are "locked" by one dominant brand.
- Feature and workflow questions are 76.6% of everything buyers ask AI, and they are the most crowded and least ownable.
- Pricing and integration questions are the most winnable (top brand near 39%, roughly a quarter locked), but they are only 7.6% of recommendations.
How we measured contestedness
We counted brand recommendations for specific topics in ChatGPT and Google AI Overviews answers from October 19, 2025 to April 25, 2026. Each brand counted once per topic per answer, even if it appeared twice. That left 612,952 recommendations across 338,580 distinct topics, 48,739 brands and 113,670 answers. We compared competition only for the 2,661 topics with at least 20 recommendations. Rarely observed topics may look uncontested simply because we have too little evidence. These results describe recommendations, not all mentions, and do not cover Perplexity or Gemini.
Features and workflows are 77% of what buyers ask AI
Before you decide which questions to fight for, look at what buyers actually ask. Parse classifies each topic by type, and the mix is lopsided. Feature requirements ("ad-free streaming," "fluoride-free toothpaste") were 43.3% of all brand recommendations. Workflow topics ("AI adoption change management," "visual sales pipeline management") were another 33.3%. Together, feature and workflow questions account for 76.6% of everything AI fields about brands. General category questions were 12.3%. The topics buyers supposedly obsess over, price and integrations, barely register: pricing and contract questions were 4.4% of recommendations, integration requirements 3.2%. Audience, business-size, and location topics were rounding errors. This matters because the volume is also where the competition is. The lane everyone assumes is the main event, "which is the best [category] tool," is precisely the lane with the most brands already standing in it.
The questions AI answers most are the hardest to win
Split contestedness by question type and the trap becomes obvious. The high-volume feature and workflow questions are crowded battlegrounds: a dozen or so brands each, a leader holding under a third, almost nothing locked. The low-volume pricing and integration questions are where a brand can actually own a slot: the leader holds close to 39% and roughly a quarter are locked, but there are far fewer of them.
| Question type | Share of recommendations | Median brands | Leading brand's share | Questions "locked" (leader ≥50%) |
|---|---|---|---|---|
| Feature requirement | 43.3% | 11 | 30.1% | 11.6% |
| Workflow | 33.3% | 12 | 28.9% | 10.7% |
| General | 12.3% | 14 | 23.9% | 6.0% |
| Pricing & contract | 4.4% | 8 | 38.7% | 25.8% |
| Integration requirement | 3.2% | 10 | 38.7% | 23.1% |
Read down the table and the strategy inverts: the more a question type gets asked, the harder it is to lead. A pricing or integration question is roughly twice as likely to have a dominant brand as a feature question, and it has fewer competitors to displace. Chasing your biggest feature keyword puts you in the crowdedest part of AI search; building an answer the model trusts for "[your product] pricing for 50 seats" or "[your product] Salesforce integration" puts you somewhere a recommendation can stick. The crowded lanes are not worthless, but they reward patience, and the open lanes reward speed.
Which industries are wide open, and which are locked up
Contestedness also splits hard by industry, and the gap is large enough to change where a brand spends. Service categories are knife-fights: in Professional Services, the median buyer question drew 12 brands and only 1.6% of questions had a dominant leader. Biotechnology, Transportation, Consumer Goods, and Manufacturing all ran a median of 13 brands per question. The opposite end is genuinely ownable. Consumer Electronics had a median of 5 brands per question, the leader held nearly half of recommendations, and 52% of its questions were locked.
| Industry | Median brands / question | Leading brand's share | Questions "locked" |
|---|---|---|---|
| Professional Services | 12 | 23.2% | 1.6% |
| Transportation | 13 | 25.2% | 8.1% |
| Financial Services | 11 | 27.1% | 10.6% |
| Software | 9 | 32.1% | 14.0% |
| Food & Beverage | 8 | 38.7% | 27.3% |
| Collaboration | 7 | 41.9% | 25.8% |
| Consumer Electronics | 5 | 49.8% | 52.2% |
The practical read: in a locked-up category like consumer electronics, displacing the leader is a multi-quarter project, but defending a lead you already hold is realistic. In a wide-open category like legal services or professional services, no incumbent is safe, and a focused brand can climb fast because no one holds the question. The same AI-visibility budget buys a very different outcome depending on which side of this table your category sits on.
What an owned buyer question actually looks like
The rare locked questions are instructive. They are narrow and specific, and one brand has become the near-default answer for them: "visual sales pipeline management" drew nine brands but the leader held 81% of recommendations; "design system documentation" sat at 84%; "small business help desk" at 81%. Compare those to the most contested topics we measured, which were lead-generation categories where dozens of brands get named for the same broad job. "Slip and fall legal representation" surfaced 176 distinct brands with the leader at 26%, and "custom engagement ring design" surfaced 58 brands with the leader at 11%. The lesson is not that workflows are always winnable; it is that AI crowns a leader when a brand becomes synonymous with one tightly-scoped job. Owning "the visual pipeline CRM" is achievable. Owning "best CRM" is not. The questions worth targeting are the specific ones your competitors treat as too small to bother with, including the buyer questions no brand consistently owns yet.
How to turn a contestedness read into a plan
Parse tracks AI visibility across ChatGPT and Google AI Mode, covering a public index of more than 4.7 million AI responses, 603,000 brands, and 57 million citations. A contestedness read turns that index into a target list in three moves. First, separate the questions you can lead from the ones you can only join: pull the topics in your category, sort by how concentrated the leader is, and stop pouring effort into questions where 12 brands split the recommendations evenly. Second, weight toward the specific over the broad: the narrow workflow, pricing, and integration questions are where a recommendation actually sticks, even though they get asked less. Third, defend what you already lead, because a locked question is a churning one: only 30% of brands stay visible from one answer to the next. Parse's market pages show which buyer questions your category competes on and where the field is concentrated versus open, and its brand pages map your position against named rivals on each. For the why-behind-the-pick, our breakdown of how ChatGPT decides which brands to recommend covers the mechanics, and does AI search favor big brands covers who tends to win the crowded questions.
If you want to see how AI engines describe your own brand, run a free brand check.