What changed · Search
Your buyer decided before they ever reached you
For twenty years the first thing a buyer did was arrive. Now the first thing they do is ask, and get an answer that may not include you.
By Prasann Garg, Co-Founder & CTO · Updated 16 September 2026 · 6 minute read
The consideration stage has moved off your property. A buyer with a question now asks an assistant, receives a paragraph naming two or three brands, and only then goes looking, usually for one of the names they were given, and where they land first is rarely your site: a review listing, a comparison page, a community thread. This means the most consequential moment in the purchase no longer produces a session, a referrer or a bounce. Your analytics cannot see it, your attribution model has no row for it, and the brands that lose there lose invisibly. That is a strange position for an industry that has spent two decades insisting everything must be measurable.
This is not only a website problem
It is easy to read this as a website problem, but it applies hardest where the buyer never reaches your site at all. The person who used to ask a peer, or work through a category page weighing options, now arrives having already been handed two vendor names. Either way the decision was made somewhere you have no visibility, and either way the only defense is being one of the names.
The step that disappeared
The old sequence was search, scan a list, click, evaluate, decide. Every stage after the click was measurable, and marketing organized itself accordingly: landing pages, session recordings, funnels, conversion rate optimization. An entire discipline grew around improving what happens once someone arrives.
The new sequence starts with a question and an answer. The evaluation that used to happen across five open tabs now happens inside a paragraph somebody else wrote. By the time a visit occurs, the shortlist already exists and you are either on it or you are not.
Why this is uncomfortable
Because the industry has spent a decade arguing that anything unmeasurable is not worth doing. That argument was always partly self-serving, since it favored the channels that reported well over the ones that worked. It is now actively dangerous, because the stage that decides most purchases reports nothing at all.
A brand that only funds what it can attribute will systematically underfund the moment that matters most. It will keep optimizing the second step while losing at the first, and its dashboards will look fine throughout.
The measurement you can actually do
You cannot measure the answers you were absent from through analytics, because absence generates no data. You can measure it directly, by asking the questions your buyers ask and recording which brands get named. That is a survey, and no tracking pixel will stand in for it, and the industry is going to have to get comfortable with that again.
What this changes about content
It inverts who the audience is. A page written to persuade a visitor assumes a visitor. A page written to be quoted assumes a model reading on someone else's behalf, extracting a passage, and attributing it. Those are different documents.
The second kind is less fun to write. It answers the question early instead of building to it, it prefers a number to an adjective, and it names the brand inside the useful paragraph. Most brand writing does the opposite of all three.
The part nobody wants to hear
Being named in an answer is closer to earning a mention in a review than to buying a placement. It depends on being genuinely worth citing, which is not something a campaign can produce in a quarter.
That is a real shift in what marketing budget can buy. Money still buys reach, and it no longer buys the shortlist. The brands that do well in this channel will be the ones with something worth saying, which is a slightly moralistic conclusion that happens to be true.
For twenty years the answer to "how do we get considered" was a budget question. It is becoming a substance question, and a lot of marketing was not built for that.
If you are an agency
This reframes what a client is buying. If consideration now happens off-site, the deliverable is not a landing page, it is being the source that gets quoted. That is closer to earned media than to performance marketing, and agencies with a PR heritage are better placed for it than the ones that grew up in the auction. It also needs a different reporting conversation, because the honest metric is which brands get named, and session conversion answers a different question.
How we do this ourselves
This is the problem the AI Search Visibility exists for: asking the questions your buyers ask, recording who gets named and which sources were used, and tracking whether that changes.