Most Brands Are Invisible in AI Search

TL;DR

Across the markets we track, AI visibility is lopsided. In some categories one brand holds 80 to 85% of AI share of voice; in others the leader sits below 30% and the shortlist is still forming. For most companies this isn't a reputation problem, it's an absence problem. The fix is being structured, cited and present on the sources AI trusts, before your competitors are.

We track AI share of voice across the B2B categories our clients compete in, and the pattern is more lopsided than most people expect. In several of those categories, one brand takes the majority of the answers, and everyone else splits what's left.

This matters because more buyers are getting answers from AI before they ever open a website. Google now shows AI Overviews across a significant share of searches, particularly longer and informational queries. And when an AI summary appears, people click through to a traditional result far less often: Pew Research found users clicked a Google link on 8% of those visits, against 15% when no summary showed, roughly a 47% drop in click propensity (Pew Research, 2025). ChatGPT, Perplexity and Gemini go further: they can answer the question directly and surface a shortlist of brands, so businesses that aren't mentioned can be left out before the buyer ever reaches a website.

So the question that decides whether you get the customer is blunt. When someone asks AI what to buy in your category, does it name you?

What our tracking shows

Across the markets we monitor, AI visibility is rarely spread evenly. In some categories, one or two brands own the answers. In others, the field is far more open.

Our tracked data shows a category leader can hold as much as 85% of AI share of voice, while in more competitive categories the leading brand sits below 30%.

Here is how the markets compare:

Category	Leader's AI share of voice	Avg. sentiment
Construction software	85%	59 / 100
Financial advice	80%	58 / 100
Buyers agents*	48%	61 / 100
Financial advisers*	42%	59 / 100
Mortgage brokers*	40%	58 / 100
Law firms*	38%	57 / 100
B2B SaaS*	35%	61 / 100
Dental*	32%	63 / 100
Builders*	30%	62 / 100
Caravan dealers*	28%	61 / 100
Returns / post-purchase software	26%	62 / 100
Furniture*	24%	64 / 100
Jewellery*	22%	65 / 100

* Modelled estimate. Direct tracking is expanding into these categories.

At one end, construction software and financial advice are heavily concentrated. The leading brand holds around 80 to 85% of AI share of voice, and the next names barely feature. If you compete in one of these markets and AI isn't naming you, you're fighting for a very small piece of the remaining visibility. AI is running the shortlist without you on it.

Professional services look less concentrated. Buyers agents, financial advisers, mortgage brokers and law firms tend to have several brands competing for visibility, especially as location and specialisation change the answer.

Then there are the more open markets. Returns and post-purchase software has a leader at only about 26%. Furniture, jewellery and other fragmented categories spread visibility across a much larger group of brands.

That gap is the opportunity. A market where one brand already owns 85% of AI visibility is a very different problem from one where the leader owns 25 to 30%. In the open one the shortlist is still forming, and there's real room to become a name AI keeps recommending.

How we measure. We use Peec AI to run a fixed set of buyer questions daily across ChatGPT, Google AI Overviews, Gemini and Perplexity, then record which brands each answer names. Share of voice measures how consistently a brand appears across those tracked AI answers. The unmarked rows above come from our own client tracking. Asterisked rows are modelled estimates as we expand direct tracking into those categories.

Sentiment: it's presence, not reputation

Look at the sentiment column. AI describes the brands we track neutrally to positively, mostly in the high 50s to mid 60s out of 100, and that holds whether a category is concentrated or open. So for most companies, low AI visibility isn't a reputation problem. AI isn't saying bad things about you. It just isn't saying your name.

That's a different problem, and an easier one. You're adding presence, not repairing damage. Nobody has to be talked out of a bad opinion of you first.

Why absence costs you

Being skipped costs you before the buyer knows you exist. If AI names a competitor for "best [category]" and leaves you out, you've lost the click, the shortlist spot and often the sale, without ever getting the chance to compete on merit.

And it compounds. The brand AI names today is the brand it has the most material to draw on tomorrow, so the leader's advantage widens on its own. The longer you're absent, the more you're catching up to a moving target.

What gets a brand named

From our own tracking and the wider research, three things do most of the work.

Structure. AI pulls passages, not whole pages. Content that leads with a direct answer, uses headings that match the question, and puts comparisons in tables gets extracted far more often than a wall of prose. Write each key point so it stands on its own, because a standalone passage is the unit AI lifts.

Evidence. This is the biggest lever, and the most under-used. In the Princeton and IIT Delhi GEO study, adding quotations lifted how often AI pulled a page by about 40%, and adding statistics by about 37%. Citing sources helped too, but by less. The lesson holds up in practice. "We're the best" gets ignored. "Our customers cut returns by 30%" gets quoted.

Off-site presence. Most of what AI says about a brand comes from somewhere other than the brand's own website. Across B2B categories, third-party sources make up the large majority of citations, often 80% or more (AirOps offsite-signals report; Aleyda Solis AI-search citation analysis). AI leans on Wikipedia, Reddit, YouTube and review sites like G2 and Capterra. If you're not on those, you're close to invisible to the models, however good your own pages are.

Where to start

The moves that pay off first:

• Build genuine comparison pages, the "[you] vs [competitor]" and "best [category]" pages buyers actually search. Make them real comparisons, not sales pages in disguise.
• Put dated, sourced numbers on your key pages. A figure with a date and a source is the kind of line AI quotes.
• Earn mentions on the third-party sources AI trusts in your category, the review sites, forums and roundups that keep turning up in answers.
• Check your robots.txt isn't blocking the AI crawlers (GPTBot, PerplexityBot, ClaudeBot, Google-Extended). Block them and those tools literally cannot cite you.
• Keep your best pages fresh and dated. AI weights recency, so a page last touched two years ago starts a step behind.

For the underlying mechanics on each of those levers, see the guide to getting cited by ChatGPT and Perplexity and the primer on generative engine optimisation.

The window is still open

The brands sitting near the top of AI share of voice today didn't get there on product alone. They were structured, cited and present on the sources AI trusts before anyone else bothered. That lead compounds, so the gap widens the longer you leave it. In the more open categories it isn't sealed yet, but it settles a little more every month someone else moves first.

If you want to know where you stand in AI answers right now, and the two or three moves that would shift you fastest, that's what we measure. Start with a free AI visibility audit and we'll show you your share of voice by category, and where the fastest gains are. Senior-led, two business days, no obligation.

Frequently asked questions

How do I check whether AI names my brand?

Run your top 15 to 20 buyer questions through ChatGPT, Perplexity and Google, and record whether you're named and who else is. Do it a few times, since answers vary. Tools like Peec AI track it daily across engines at scale, which is how we pulled the numbers here. For a quick check on one query, try our free AI Visibility Checker.

Is low AI visibility a reputation problem or a presence problem?

For most companies, presence. In our tracking, AI talks about brands neutrally to positively (average sentiment in the high 50s to mid 60s out of 100). The issue usually isn't what AI says about you, it's that AI doesn't mention you at all.

Does traditional SEO still matter for AI search?

Yes. Good SEO is the base. AI search adds structure, evidence and third-party presence on top. You need both, not one instead of the other.

Which sources do AI models cite most in B2B categories?

Third-party ones, mostly: review sites like G2 and Capterra, Wikipedia, Reddit, YouTube and independent roundups. The exact mix shifts by category, and it's one of the things our tracking pins down for each client.

How long does it take to improve AI share of voice?

It varies by category and starting point. New comparison pages and off-site mentions tend to show up in AI answers over a few months rather than weeks, because the models need to re-crawl and re-weight the new material.

Which AI engines should I be tracking?

Start with the ones your buyers use: ChatGPT, Google AI Overviews, Gemini and Perplexity. Track them together, because share of voice can differ a lot between engines, and a brand that wins in one can be absent in another.