The marketing game did not change. The decision engine did.
People no longer search and decide in one motion. They discover, compare, validate, and only then commit. That journey now runs across platforms, creators, communities, and AI answers, and a brand can be eliminated at any stage without ever knowing it was in the running.
What actually changed in how people decide?
Google is moving toward answer-led search. A question that used to return ten links now returns a written answer with a few sources attached. Gemini cites roughly 6.6 sources per answer and names about 4.4 brands. If you are not one of them, you are not in the consideration set.
TikTok and Instagram have become places people go to find things, not only to be entertained. Pinterest tends to show demand before it trends, which makes it one of the few forward indicators available for free. And creators now shape what an audience believes, not only what it has heard of. A recommendation from someone already trusted does work an ad cannot buy.
Why does being cited matter more than ranking?
The answer box is where the decision happens now. When a generative engine writes the summary, the brands named in it get considered and the rest get nothing, wherever they sit on the results page underneath.
Visibility in one engine also does not carry over to another. Ahrefs studied 76.7 million AI Overviews alongside roughly 1.9 million prompts across ChatGPT and Perplexity. Of the 50 most-mentioned websites, only seven appeared in all three. That is a 14 percent overlap. Doing well in ChatGPT tells you very little about whether Gemini will mention you.
What decides whether an AI engine cites you?
Mostly the plumbing, not the prose. Yext analysed 17.2 million AI citations across ChatGPT, Perplexity, Gemini and Claude and found that visibility is governed by retrieval logic rather than content quality alone. Verified, structured, directly distributed data accounted for 54.53 percent of distinct citation sources.
So more than half of what gets cited is information a machine can confirm: company registrations, business profiles, listings, marked-up data. Not essays.
Content still has a lever. Princeton's research on generative engine optimisation found that citing sources, adding statistics and including direct quotations improved AI visibility by 30 to 40 percent against unoptimised content. That works on top of the retrieval layer though, not instead of it.
Where should a brand spend first?
Start with the record rather than the writing. Check that the business appears consistently everywhere a machine treats as authoritative: the company registration, the business profile, the social accounts, the directory listings. Conflicting or missing records split one brand into several half-entities, and none of them build up trust.
Once that holds, make the content quotable. Lead with the question someone would actually type, answer it in the first two sentences, attach a number, and say where the number came from. Attribute opinions to a named person whose record can be checked.
It also helps to be where the engines already look. Roughly 43 percent of citations across Google's AI search products point back to Google's own properties, including YouTube, Maps and the Knowledge Graph. YouTube alone accounts for 29.5 percent of AI Overviews citations. A brand absent from all of them is invisible across nearly half the citation surface before anyone writes a word.
What does this mean for 2026?
Growth will not go to the loudest brand. It will go to the one that is easiest to verify.
"Audiences no longer take one path to a decision," says Mariam Tania, Founder of KATA Consulting. "They gather proof from several places before they commit, and most of those places are not owned by the brand. Our job is to make sure the brand is consistent and verifiable everywhere that checking happens."
For most brands that means the dull work comes first. Fix the registration, claim the profile, make the numbers checkable. The writing gets easier once the record underneath it is straight.
