Something Inc.Schedule a free consultation
GEO

Topical authority in AI search: why 89% of demand has no owner

Kevin Indig ran 1,094 categories through ChatGPT and found only 15% have a brand that consistently wins the answer. Here's what's happening to the other 85%.

JBJosh BernsteinManaging Partner · JUL 24, 2026 · 9 MIN READ
15.2%
of categories have a clear AI-search owner
89.3%
of estimated demand sits in categories with no owner
90.4%
of owners retain the #1 spot month over month
69.9%
citation rate for the single most-mentioned brand in a category
THE HEADLINE FINDINGKevin Indig tested 1,094 U.S. categories against ChatGPT and found a clear, stable leader in only 15.2% of them. The rest of the market, 89.3% of the estimated demand, is being answered by whoever ChatGPT happens to name that day. Most brands are competing for a prize almost nobody has claimed yet.
TL;DR · 60 SECONDSGrowth Memo's July 2026 study ran 1,094 categories through five ChatGPT prompts each, tracking 220,000+ domains and 600,000+ citations. Only 15.2% of categories have a stable owner. Owners that do exist are hard to dislodge, retaining the top spot in 90.4% of month-over-month comparisons, and they carry 55.7% higher branded search volume than the runner-up. The other 84.8% of categories are wide open, which means most of the market is still winnable if you move before a rival locks it down.

Every GEO conversation eventually turns into a fight over one brand's citation rate. That's the wrong fight. The bigger number is how much of the category has no stable citation leader at all, and it's most of it.

Kevin Indig published the study on Growth Memo on July 20, 2026, and Search Engine Land syndicated the full write-up the same week. The design was straightforward: take 1,094 categories relevant to U.S. B2B and consumer buyers, run five ChatGPT prompts against each one, and track which domains got cited across 220,000-plus tracked domains and 50,000-plus brands. Between January and June 2026, that produced more than 600,000 logged citations. Indig then defined "ownership" as a brand holding the top citation position consistently across the study window, not just showing up once.

That scale is what makes the finding hard to dismiss as noise. This isn't a spot-check of a dozen categories where a fluke result could swing the headline number. It's six months of repeated prompting across more than a thousand distinct markets, wide enough to cover everything from enterprise software categories with a handful of serious competitors to consumer categories with hundreds of plausible answers. When a dataset that size still comes back with an 84.8% unowned rate, the honest read isn't that the methodology missed something. It's that most categories genuinely haven't been settled yet, and the brands assuming otherwise are working off an outdated mental model of how competitive their space actually is inside an AI answer.

Ownership here isn't the same thing as a single citation. Plenty of brands get named once in an AI answer and never again; that's noise, not authority. Indig's definition requires a brand to hold the top position across repeated prompts and repeated months. That's a much higher bar, and it's why the ownership rate comes out so low: 15.2% of categories clear it, meaning 84.8% of categories never produce a brand that consistently wins the answer.

This matters because most teams measure the wrong thing. A blended mention-rate number can look healthy in a category where no one, including you, has actually locked anything down. The real question isn't "did we get cited this month," it's "is anyone building a moat here, and is it us." In the 84.8% of categories without an owner, the answer to the second half is almost always no.

The data: 600,000 citations, 1,094 categories

The ownership rate isn't flat across category sizes. Indig found that higher-volume categories, the ones with the most estimated AI-search demand, are owned less often than smaller ones: 11.3% of high-volume categories have a stable leader, against 19.0% for lower-volume categories. That's the opposite of what most GEO teams assume. The biggest prizes are the least claimed, probably because they're the hardest to dominate and the most crowded with plausible answers.

High-volume categories11%
Low-volume categories19%
Blended average (all 1,094)15%

Category ownership rate by demand tier (Growth Memo, Jul 2026)

The categories that do have an owner show a wide, measurable gap between the leader and everyone else. Owners carry meaningfully higher branded search volume, organic traffic, and independently scored authority than the next brand in the same category, which is the closest thing to proof that owning the AI answer and owning the underlying market position are the same behavior measured two different ways.

METRICCATEGORY OWNERRUNNER-UPOWNER'S EDGE
Branded search volumeBaselineBaseline − 55.7%+55.7%
Organic trafficBaselineBaseline − 48.4%+48.4%
Independent authority scoreBaselineBaseline − 52.5%+52.5%
Month-over-month #1 retention90.4%n/a

Why owners stay owners

Once a brand takes the top spot in a category, it rarely loses it. Indig's month-over-month comparisons show a 90.4% retention rate for the incumbent leader, which means the leaderboard is close to locked in the categories that have one. That's consistent with how our own 2026 citation study reads engine behavior more broadly: engines default to sources they've already trusted and cited successfully, so the first mover in a category gets a compounding advantage that's expensive for a challenger to unwind.

There's a practical implication buried in that retention number that most brands miss: the 15.2% of categories with an owner are, for all purposes, closed for new entrants this year. Displacing a brand with a 90.4% month-over-month retention rate and a 55%-plus edge on branded search and authority isn't a content problem you solve with three new blog posts. It's a multi-quarter repositioning effort, and most challengers underestimate the cost of trying. The far cheaper move, and the one this data actually argues for, is redirecting that same effort toward the 84.8% of categories where no incumbent exists yet to displace.

1Correlation is weaker than you'd guessThe relationship between raw mention count and citation rate is actually slightly negative (−0.229) across the dataset. Being mentioned constantly doesn't guarantee you're the one getting cited as the source; the two numbers move somewhat independently.
2But the top mention still wins most of the timeThe single most-mentioned brand in a category still gets cited 69.9% of the time. Volume alone isn't sufficient, but it's still the strongest single predictor available.
3Ownership compounds through search behavior, not just contentHigher branded search volume feeds the model's sense of who the category belongs to, which is a signal you build over months of demand generation, not one blog post.
Most brands are optimizing content for a citation fight that's already been decided in someone else's favor, while ignoring the much larger set of categories where the fight hasn't started.

Where the unowned 89% is going

If 84.8% of categories have no stable owner, the citations in those categories are still happening, they're just scattered. ChatGPT answers the question every time it's asked; it just answers it with a different brand, or a directory, or a comparison site, depending on the day and the prompt. That volatility is itself the opportunity: a brand that shows up consistently, even without full ownership yet, starts pulling ahead of competitors who show up sporadically. Consistency is a variable you can control long before you can control outright ownership.

This is also where format matters. Across the categories we track for clients, comparison and alternatives content is disproportionately represented among the brands starting to consolidate a fragmented category, because it's the format that answers the buyer's actual next question instead of restating the category definition. A scattered category isn't evidence that GEO doesn't work there; it's evidence that no one has built the content that would let an engine settle on an answer.

There's also a timing signal worth watching inside the negative citation-mention correlation Indig found. A slightly negative relationship between raw mentions and citation rate (−0.229) means a category with rising mention volume but no consolidated citation leader is actively unsettled right now, not stuck in a permanent muddle. That's a different situation than a mature category that's been fragmented for years; it's closer to a market mid-consolidation, which is exactly the window where a deliberate push, the comparison content, the branded demand generation, the third-party corroboration, has the highest odds of tipping the category rather than just adding another voice to the noise.

How to claim a category before someone else does

Step 1
Run the ownership test on your own categoryPrompt ChatGPT, Perplexity, and AI Mode with your buyer's actual questions five separate times over two weeks. If the cited brand changes each time, you're in the 84.8%, and it's open.
Step 2
Build the comparison page firstUnowned categories get settled by whichever brand gives the engine the cleanest answer to "what are my options," not whichever brand has the most content overall.
Step 3
Feed branded search volume deliberatelyIndig's data ties ownership to branded search volume, not just citations. Demand generation and PR that drive people to search your name directly reinforce the same signal engines use to decide who owns the category.
Step 4
Re-test monthly, not quarterlyOwnership calcifies fast, 90.4% retention once it locks. If you wait a quarter to check, you may be measuring a category that already has a new incumbent.

The takeaway from Indig's data isn't that AI citation is a lost cause for latecomers. It's closer to the opposite: most categories are still up for grabs, and the brands treating this as a settled game are the ones about to lose it to someone who didn't. Run the ownership test on your three highest-value categories this week. If none of them show a stable leader, that's not a gap in the data, that's your opening, and it's the same gap our GEO engagements are built to close before a competitor notices it's there. We ran exactly this play on a B2B security platform that went from scattered mentions to a top-three citation position in its category within two quarters, then wired the result into a reporting model that tracked ownership, not just visibility, every month after.

See where you are cited today

A free snapshot audit of your rankings and AI citations before we ever talk.

JB
Josh BernsteinMANAGING PARTNER, SOMETHING INC.

Josh leads work at the intersection of SEO and generative engines at Something Inc., helping B2B brands get ranked and cited across every major AI engine.

Free consultation

Let us be the last SEO agency you ever work with

A 30 minute call and a free audit of your SEO and GEO position. You keep the findings either way.