90% of brands have zero AI mentions. That’s the opportunity.

A new study landed yesterday from Victorious, syndicated through Search Engine Journal, that's going to get framed as bad news. The headline number is that 89.8% of brands have zero AI search mentions across the eight major platforms — ChatGPT, Perplexity, Gemini, AI Overviews, AI Mode, Copilot, Claude, Meta AI. Of 177 brands tested across healthcare, SaaS, finance, e-commerce and legal, only 18 registered any mentions at all.

The industry reaction will be predictable. Panic merchants will use it to sell GEO retainers. The "AI search is overhyped" crowd will use it to argue nothing's happening. Both readings miss what the data actually says.

The real signal in that study is the opposite of urgency. It's that for the overwhelming majority of UK businesses reading this, the AI search race they keep being told they're already losing hasn't actually started in their vertical. The competitive field is small. The incumbents are few. And the levers that move the needle are the ones competent SEOs have been pulling for fifteen years.

That's a much more useful framing than the headline number suggests. Let me unpack it.

What the study actually measured

107,011 AI responses. Eight platforms. Q1 2026 data. Brands cross-referenced against Semrush organic data and Authority Score. Mention rate and citation rate tracked separately, which is the right call — being named in an answer and being linked as a source are two different outcomes with different drivers.

The study is sponsored content. Worth flagging. Victorious is selling SEO services and the conclusions conveniently support buying SEO services. But the methodology is reasonable and the dataset is large enough to draw real signal from, and the findings line up with what other research from late 2025 onwards has been showing: traditional authority signals correlate strongly with AI visibility, and most "GEO-specific" interventions do not.

The four findings, briefly:

1. 1.89.8% of brands had zero AI mentions
2. 2.Patterns vary substantially by vertical
3. 3.Healthcare, SaaS and financial services brands get mentioned *and* cited consistently
4. 4.Authority Score correlates with AI visibility (i.e. the SEO fundamentals still matter)

That last one is the bit nobody's quoting on LinkedIn this morning.

Why "90% have no AI mentions" is not the disaster it sounds like

If you read that number as "AI search is failing brands," you've got it backwards. AI search isn't failing brands. Most brands haven't done anything that would make them visible to AI search yet, and AI search is responding accordingly.

The race hasn't started in most categories. It's just that nobody told the people writing the urgency-pitch decks.

This is the inversion most commentary is going to miss. The platforms aren't broken. They're working exactly as you'd expect a citation engine built on top of authority signals to work — they're surfacing the small number of entities that have established themselves as credible sources in each vertical. Everyone else doesn't appear because there's no reason for them to appear.

The race hasn't started in most categories. It's just that nobody told the people writing the urgency-pitch decks.

Compare this to organic search circa 2008. If you'd surveyed 177 random small businesses, the vast majority would have had no meaningful Google visibility either. Not because Google was broken. Because they hadn't done the work yet. The companies that did the work in the next five years built compounding moats. The ones that waited until 2015 to start are still trying to catch up.

That's the actual shape of the opportunity here. Not "everyone needs a GEO retainer yesterday." More like "the table is being set, and most of the chairs are still empty."

The verticals where it's already started

The study breaks this out cleanly. Healthcare, SaaS and financial services are where AI platforms already have stable, repeatable answer patterns. The reasons aren't mysterious:

• Healthcare has clean entity identifiers — registered practice names, location data, specialty codes, professional body affiliations. AI platforms can resolve "best private GP in Manchester" because the entity graph is well-defined and reinforced by directories, review sites and professional registries.
• SaaS has comparison content everywhere. Listicles, alternatives pages, Reddit threads, G2 reviews. The training data is saturated with structured opinions about which tools do what.
• Financial services has regulatory disclosure requirements that produce predictable, schema-friendly content, plus heavy editorial coverage from a small number of trusted publishers.

If you operate in one of these three, the competitive field for AI visibility is already forming. The incumbents are probably the same entities that dominate organic — because the signals overlap. If you're not one of them, you're competing for the cited-but-not-mentioned tier, or the mentioned-but-not-cited tier, depending on what your content actually does.

If you operate in e-commerce/retail or legal services — the two verticals where the patterns were messier — the field is much more open. That's good news, if you read it correctly.

The Authority Score finding is the real story

Buried under the 90% headline is the actual lesson: brands with higher Authority Scores were more likely to be both mentioned and cited. The traditional SEO signals — domain authority, backlink quality, editorial coverage — predict AI visibility.

This is the third or fourth major piece of research saying the same thing in the last six months. Princeton's GEO study. The various log-file analyses out of the agency world. Now Victorious. The consistent finding: the things that get you cited in AI answers are mostly the things that got you ranked in organic. Brand authority, real backlinks from real publishers, structured content, technical hygiene.

What doesn't show up as a strong predictor: any of the GEO-specific tactics being aggressively sold right now. No correlation with llms.txt files. No correlation with chunking experiments. No correlation with the various "AI-optimised" content rewrites.

That doesn't mean those things never matter — Mike King's argument that Google's guidance is self-serving is also correct, and his case for JavaScript-based content protection from AI crawlers has technical merit. But the load-bearing work for AI visibility, for the overwhelming majority of businesses, is the same work it has been for years. Done properly. Sustained over time.

What this means if you're a UK business reading the headlines

A few things to take from the study that won't be in most write-ups:

You probably haven't missed the boat. If you operate outside healthcare, SaaS or finance, the competitive field for AI visibility in your vertical is still forming. Acting now puts you in front of the wave, not behind it.

The work to do is the work you already knew about. Build real authority. Earn editorial coverage from publications your customers actually read. Tighten technical foundations. Use schema where it's genuinely useful. Publish content that's worth citing because it says something other content doesn't.

The measurement situation is still bad. None of the public tools give you a clean view of how AI platforms are treating your brand, and the "AI citation tracking" tools being sold right now are mostly the wrong abstraction. Manual prompt-testing across the main platforms is genuinely useful and almost free. Do it monthly. Note what changes.

Be sceptical of urgency pitches. If an agency is telling you that your category is "already lost" to AI-savvy competitors and you need a six-figure GEO programme to catch up, ask them which competitors specifically. In most UK verticals outside the three named above, that's a sales narrative, not a market condition.

The honest limit

One caveat. The study tested fixed vertical-specific prompts. Real users ask weirder, longer, more specific questions than benchmark prompts. The 89.8% zero-mention figure is a benchmark, not a complete picture of how brands appear in the long tail of actual AI queries. Some of those brands almost certainly get surfaced in narrower queries that the study didn't test for.

That cuts both ways. It means the gap might be smaller than the headline suggests for some brands. It also means it might be larger for others. Either way, the central reading holds — the AI search field is still forming in most verticals, and the work that wins in it is mostly the work that won in organic.

The race hasn't started for most of you. That's the actual news. The trick is to start moving while everyone else is still arguing about whether the gun's gone off.