Most ChatGPT topics still have no owner. In fintech and crypto, that's the opening.

Semrush partnered with Kevin Indig and Growth Memo to track 1,094 topic categories across six months of ChatGPT data, over 220,000 domains, 50,000 brands, and 600,000 citations. It's the largest look yet at how brand visibility in ChatGPT actually behaves at the category level, rather than the single-prompt level most rank trackers still measure. The full study is worth reading, but three findings matter especially for fintech, crypto, and Web3 companies that don't have a 20-year head start on domain authority.

The topic is the unit, not the prompt

The study defined a category owner as a brand that shows up in at least four of five related prompts, definition, comparison, alternatives, use case, and buying question, with at least a five-percentage-point lead over whoever's in second place. Miss that bar and you're either an "emerging leader" or the category is still wide open.

Only 15.2% of the 1,094 categories tracked had a clear owner by that definition. 53.7% had no brand appearing in even three of five prompts at all. That maps almost exactly onto how someone actually researches a fintech or crypto purchase: what is it, how does it compare, what else is out there, does it fit my use case, and how do I actually buy or use it. Ranking for one of those five questions and assuming you're covered is the mistake. The study found that assumption is usually wrong.

The biggest categories are the least owned

This is the part that should reframe how fintech and crypto teams prioritize. The researchers split their 1,094 categories by search demand and found clear ownership in only 11.3% of the highest-demand topics, versus 19% of the lower-demand ones. And the high-demand half accounted for 98% of all the AI search volume in the sample.

In other words: almost every question that actually matters, the ones with real buyer intent behind them, lands in a category where no single brand has locked things down yet. In crypto and Web3 specifically, where the biggest names (the Coinbases, the Binances) have obvious dominance on broad terms but can't possibly own every sub-category, DeFi lending, specific L2 ecosystems, cross-chain infrastructure, stablecoin yield, that gap is even more pronounced. The same is true in fintech verticals like trade surveillance, cross-border payments infrastructure, or embedded finance, categories too narrow for the household names to have fully claimed, but with real commercial intent behind the searches.

Domain authority barely predicts who wins

Here's the finding that matters most if you've been told AI visibility is just SEO with extra steps. The study compared category owners against runners-up on three classic SEO signals: branded search volume, organic traffic, and Authority Score. Owners only had the edge on those metrics about half the time, branded search volume 55.7%, organic traffic 48.4%, Authority Score 52.5%. In nearly half the categories, the brand with the stronger domain-level SEO profile still lost the topic to a competitor.

That's consistent with what I've argued about AEO generally: AI answer engines don't inherit Google's authority hierarchy wholesale. They reward entity clarity and topical depth, being able to be described the same, consistent way everywhere the model looks, more than they reward raw backlink volume. For a fintech company competing against institutions with decades of domain history, or a crypto protocol competing against exchanges with ten times the traffic, that's the opening. You don't need to out-authority the incumbent. You need to out-cover them on the specific cluster of questions that define the category.

Strong SEO gets you into the running. It doesn't win you the topic. Depth of coverage does.

Once you own a topic, hold it, but only with a wide margin

The study's fourth finding is a warning against declaring victory too early. Category owners with a comfortable lead kept their position in 90.4% of month-over-month comparisons. But when leadership changed hands, the median lead at the time of the flip was only 1.3 percentage points, versus a 2.9-point median for the leads that held. Margin mattered more than momentum.

For a crypto project that just landed a mention in a ChatGPT answer, or a fintech brand that shows up for one comparison prompt, that's not the finish line. It's a fragile position that a faster-moving competitor can take within a month or two. The goal isn't a single good citation. It's a wide enough lead across the full prompt cluster that it's expensive for anyone else to close the gap.

What this means in practice

Three things I'd take directly from this into a fintech, crypto, or Web3 content strategy:

Map the actual prompt cluster around two or three core topics, not a sprawling calendar. For each one, build content that answers the definition, the comparison, the alternatives, the use case, and the buying question, deliberately, not as whatever the blog happened to cover that quarter.

Prioritize mentions in the sources the model actually pulls from in your vertical, Cointelegraph, Messari, protocol documentation, and community discussion for crypto; analyst reports and trade press for fintech, over generic backlinks aimed at moving an Authority Score that this data says barely matters half the time anyway.

Track topic-level share, not single-prompt rank. If you're only checking whether you show up for one branded query, you have no idea whether you actually own the category or just got lucky once.

None of this replaces the fundamentals I've written about before: entity consistency, structured data, a site that describes itself the same way everywhere. What this study adds is a number to point to. Most of the categories that matter to fintech and crypto buyers are still unclaimed, and the brand with more backlinks doesn't automatically win them. For companies that never had Google-era domain authority to begin with, that's about as good an opening as AI search is going to offer.