Only 1.5% of buying prompts name a brand. Your product gets picked in a conversation you're not in.

Pew measured the click collapsing from 15% to 8% when an AI summary appears. A separate read of 35,000 ChatGPT citations shows where the decision moved to — and it's mostly places you don't own.

Start with the one number here that isn't someone's marketing. Pew Research put browsing software on 900 US adults and watched 68,879 real Google searches in March 2025. Not a survey about how search feels. Actual behaviour.

When an AI summary appeared, people clicked a traditional search result on 8% of visits. When it didn't, 15%. Roughly half. They clicked a link inside the summary itself on 1% of visits. And they ended the browsing session entirely on 26% of pages with a summary, against 16% without. An AI summary showed up on 18% of searches at the time; that share has only gone one direction since.

So the click is going. That part is real and measured, and if your plan for getting found was "write useful things and rank," the ground under that plan is moving.

Now the part where I'd ask you to slow down, because there's a story being sold about what replaces it.

You've seen the claims: AI referral traffic converts 5x better, 9x better, it's the highest-intent channel anyone's ever seen. Almost every one of those numbers comes from a company selling AI-visibility optimization. The largest independent look I can find points the other way — 973 ecommerce sites, $20 billion in revenue, 12 months of first-party analytics, more than 50,000 ChatGPT-driven transactions compared against 164 million from traditional channels. ChatGPT referrals were about 0.2% of sessions, roughly 200x smaller than Google organic. They converted about 13% worse than organic search, and worse than affiliate by a wide margin. Only paid social did worse. The trend line is up and to the right, and the authors project no parity with organic search within a year. Another study, from Amsive, found no significant conversion difference at all and LLM referrals under 1% of sessions.

Hold both: the old channel is shrinking, and the new channel is currently a rounding error that converts fine, not magically. Anyone selling you urgency on this is selling you something.

Which leaves the actual question. If the click is disappearing and the referral is tiny, where is the decision happening?

Kevin Indig published an analysis four days ago that I think is the most useful thing available on this. About 35,000 ChatGPT citation URLs, SaaS-related prompts, US, December 2025, sorted by where the buyer is in their journey. Strip out vendor domains and look at what's left: user-generated content platforms hold 17.1% of cited domains. Publishers hold 4.0%. More than four to one.

And it doesn't move. Review sites swing with intent — 7.4% at discovery, up to 13.2% at evaluation, back to 8.4% when someone's down to two finalists. UGC sits at 17.8, 18.2, 15.1, 17.2 across those same stages. It's not a lever you pull at the bottom of the funnel. It's the floor the whole thing rests on.

But here's the number I actually want to hand you. More than half the prompts in that set used commercial language — people actively trying to buy something. Only 1.5% of them named a vendor brand.

Sit with that. The overwhelming majority of buying conversations about your category are happening without your name in them. Nobody is asking whether your product is good. They're describing a problem and asking what exists. By the time a name enters the conversation, the shortlist was already built — out of Reddit threads, Wikipedia, LinkedIn posts, and forum comments. Those three platforms alone are 99% of the UGC citations in the sample, and Wikipedia by itself carries 10 to 14 points of that 17-point floor.

For an indie maker that reframes the job. You are not being evaluated. Your category is being described, and you're either legible inside that description or you're absent. Brand marketing — the thing that makes people type your name — is the later problem. The earlier one is whether the plain-language description of what you do matches how a person in trouble would actually phrase it.

Two honest limits before anyone builds a strategy on this. One engine, one country, one month, and 91% of citations appear in only a single engine, so ChatGPT results are not AI search results. And the platforms are volatile: Reddit's AI mentions fell 10.9% in the 28 days to June 8, then rose 18.1% three weeks later. This is a moving target, not a channel you can lock in.

What I'd actually do, and I'll keep it short because I don't think anyone knows more than this yet:

Write the plain sentence. Not your positioning — the sentence a person would use to describe their problem to a friend. Make sure it exists somewhere public, in your own words, in full.

Be present in the places your category gets discussed, as a person answering questions, not a brand doing outreach. This is slow and it doesn't have a dashboard.

And stop reporting on brand-name search volume as if it's your discovery metric. It's measuring the 1.5%.

There's an obvious awkwardness in me writing this here, on a forum, where posting is the exact behaviour I'm describing — so I'll just name it rather than pretend. I post here because it's where the people I want to talk to actually are, and I'd be writing this the same way if none of it were ever cited.

Murror sits badly inside all of this, which is why I've been thinking about it. Nobody searches for an AI companion for understanding your emotions. They type something like "why do I keep having the same fight." The gap between the category name and the sentence a real person uses is enormous, and that gap is where products in spaces like mine quietly disappear. I don't have this solved. I've just stopped assuming the name is the thing that gets found.