Split 1,094 topic categories by demand and the result runs backwards from what you would expect: the busiest ones are the least settled. Semrush's AI Visibility Toolkit team, working with independent SEO analyst Kevin Indig of Growth Memo, ran those categories through ChatGPT from January through June 2026, covering more than 50,000 brands, 220,000 domains and 600,000 citations, and published the results on July 21, 2026. Each category got five prompts a buyer would actually type: definition, comparison, alternatives, use case, buying decision. Ownership was defined tightly. One brand has to lead at least four of the five, by five points or more. Only 15.2% of categories clear that bar. Another 31.2% have an emerging leader. The other 53.7% have nobody. Among high-demand categories the clear-owner rate falls to 11.3%, against 19% in low-demand ones. And once a brand does take a category it tends to stay there: 90.4% of established owners held the top position month over month.
Two findings sit underneath that split. The signals most teams already pay to track (branded search volume, organic traffic, Authority Score) predicted poorly which brand would take a category. And brand mentions and citations aligned in only 21% of cases: showing up in ChatGPT's answer text is not the same as being the cited source.
Fact bullets:
- Of 1,094 categories analyzed (Jan–Jun 2026): 15.2% have a clear owner, 31.2% have an emerging leader, 53.7% are unsettled (Semrush, published 2026-07-21).
- High-demand categories are less settled than low-demand ones: 11.3% of high-demand categories have a clear owner vs. 19% of low-demand categories.
- Once a brand establishes ownership, it holds it: 90.4% of established owners kept the top position month-over-month.
- Brand mentions and citations aligned in only 21% of cases: appearing in the answer is not the same as being the linked source.
Credibility: High. Primary study with disclosed sample size (1,094 categories, 50,000+ brands), date range, and methodology; co-authored by an independent analyst (Kevin Indig) and reported by independent trade press (Search Engine Land). Bias to note: Semrush sells AI-visibility tracking tools and has a commercial incentive to frame AI-search ownership as high-stakes. The disclosed numbers are checkable, but the framing carries that incentive.
Sources:
Search Engine Land: "ChatGPT topic ownership is rare, and SEO alone doesn't explain it," Danny Goodwin, July 21, 2026 — https://searchengineland.com/chatgpt-topic-ownership-seo-482901
Semrush: "AI visibility is a topic-level game: A study of 50,000 brands in ChatGPT" [Study], July 2026 — https://www.semrush.com/blog/chatgpt-topic-authority-study/
Stay ahead of AI search
The changes that move your rankings and AI citations — and the exact move to make — before they cost you. Free, three times a week.
For business decision-makers tracking AI visibility, read this as a map of which categories have already been answered for, and by whom. If a competitor clearly owns yours, going at the same five prompts head-on is a slow way to spend a year: 90.4% retention is what a settled category looks like from the outside, and nothing in the data shows a mechanism for dislodging one. The better question is whether an adjacent, less-settled category is winnable instead. If your category sits in the unsettled 53.7%, the same number works in your favour. Whoever gets there first is likely to still be there next month. The demand split matters here too: the categories still sitting open tend to be the ones with buyers in them, not the leftovers nobody wanted.
The finding that will cost people quietly is the 21%. Most reporting a marketing team looks at counts mentions, and a mention is the cheaper of the two events: your brand appeared in the text. Being the cited source is what puts a link in front of the buyer and what the engine treats as the page it learned from. Those are different outcomes, and they belong on different lines of the report.
None of this retires SEO. Organic rankings still feed the pages these prompts pull from, and Authority Score still says something real about a domain. The study's claim is narrower and more useful than that: those numbers will not tell you whether you own the answer in a given topic. That is a separate measurement, and nothing on the SEO dashboard runs it for you.
This week:
- Pick your top 3–5 priority categories and have your web or content team put each one through ChatGPT using the study's five prompt types (definition, comparison, alternatives, use case, buying decision). Record how many of the five you appear in at all. Three of five is where a brand starts to look like a contender.
- Split mentions from citations in whatever AI-visibility reporting you already get, WorkDuo or otherwise. A category where you are named but never cited is a different problem from one where you are absent, and it takes a different fix.
- For any category that comes back unsettled, move its comparison, definition and alternatives content into this quarter's plan rather than next year's. The 90.4% number is the argument for doing it now.
Search Engine Land: "ChatGPT topic ownership is rare, and SEO alone doesn't explain it," Danny Goodwin, July 21, 2026 — https://searchengineland.com/chatgpt-topic-ownership-seo-482901
Semrush: "AI visibility is a topic-level game: A study of 50,000 brands in ChatGPT" [Study], July 2026 — https://www.semrush.com/blog/chatgpt-topic-authority-study/