Where the $750 Billion AI Search Number Comes From

The $750 billion figure comes from one place: McKinsey’s October 2025 article on winning in the age of AI search. Read it closely and the number gets harder to use, not easier.

On that single page, McKinsey describes the figure three ways. The summary says AI search could “impact” $750 billion in revenue by 2028. The headline statistic says $750 billion of consumer spend will “flow through” AI-powered search. The body says $750 billion in US revenue will “funnel through” it. The footnote behind all three reads, in full, “McKinsey projection.” The article itself shows no method, no inputs and no definition of what counts.

That doesn’t make McKinsey wrong. It makes the figure a headline rather than a planning input. And it is the least documented number in a report that contains several better ones.

Three verbs, three different claims

The verbs matter because each one describes a different amount of money.

“Impact” is the widest. Any purchase where AI search played some part, however small, could count. A shopper who read one AI Overview before buying a fridge in a store would qualify.

“Flow through” is narrower. It suggests spending that passes through AI search as a channel, the way we talk about revenue flowing through a marketplace.

Neither means $750 billion of new revenue, or $750 billion moving away from search ads, or $750 billion at risk for brands that ignore AI. Readings like these have appeared in articles that cite the figure, including the earlier version of this one. Without a published method there is no way to tell which reading McKinsey intended, or to check the arithmetic behind it.

This is the distinction worth getting right before a number goes into a budget deck. Influenced spend describes how many purchases a channel touches. Channel revenue describes money a channel captures. A board can reasonably ask for a plan against the second. The first is context.

Grade every number by what stands behind it

The same McKinsey article contains figures with their sources shown. They deserve different amounts of trust.

The strongest come from McKinsey’s AI Discovery Survey, fielded in August 2025 with a representative US consumer panel of 1,927 people. Half of those surveyed said they intentionally seek out AI-powered search. Among AI search users, 44 percent called it their primary and preferred source of insight, ahead of traditional search at 31 percent. These are self-reported and US-only, but the sample and date are published. You can plan around them with those limits in mind.

Next come McKinsey’s observations of AI Overview sources. A brand’s own sites often make up only 5 to 10 percent of the sources AI search references, and in consumer packaged goods and financial services more than 65 percent of sources are publishers, user-generated content and affiliate sites. The footnote cites Google AI Overview data and McKinsey analysis. The method isn’t detailed, but the finding matches independent work: a University of Toronto study found AI search engines lean heavily on earned, third-party media over brand-owned content.

Weaker is the widely repeated claim that only 16 percent of brands systematically track AI search performance. Its footnote shows a survey of Fortune 500 consumer brand CMOs with a sample of about 30. That is a conversation with thirty executives, useful as a signal, too small to describe an industry.

The forecast of a 20 to 50 percent decline in traditional search traffic for unprepared brands is footnoted as McKinsey analysis. A range that wide is a scenario band, not a forecast you can hold a channel to.

And the $750 billion sits at the bottom of the list: the biggest number, with the least shown behind it.

The useful finding is the smallest number in the report

If you take one figure from McKinsey’s article into planning, take the 5 to 10 percent. It says most of what AI search tells your buyers about you is built from pages you don’t own: publishers, reviews, affiliates, forums. That has direct consequences, and none of them depend on whether the total is $750 billion or half that.

It means your own site is necessary and nowhere near sufficient. The pages AI engines draw from for your category are a specific, findable list, and most of the ones ChatGPT cites turn out to be pages you can’t pitch at all, a pattern covered in which sites ChatGPT cites most. The rest, the review platforms and comparison articles and trade coverage, is where the work goes. How an engine picks among them is the subject of why AI search names your competitor and not you.

It also means measurement starts outside your analytics. Before you estimate what AI search is worth to you, find out where your brand appears across the sources those engines read, and where competitors appear instead. That is the job of an online presence analysis that maps your brand across third-party sources, and it answers a question the $750 billion never could: what is your share of the answers in your own category?

Your own pages still matter for the part you control. A page that states specific, checkable facts in clear sections is easier for an engine to use than one that repeats positioning, which is where search work aimed at being cited as well as ranked fits.

Three questions for the next big number

AI search statistics travel fast and lose their footnotes on the way. Before one goes into a plan, ask three things. Who measured it, and how? How many people or pages sit behind it? And what exactly does the verb claim: touched, flowed through, or captured?

The $750 billion fails the first two and blurs the third. The survey of 1,927 consumers and the 5 to 10 percent source share pass well enough to act on. A number that shows its work is worth more than a bigger one that doesn’t.

Frequently Asked Questions

Where does the $750 billion AI search figure come from?

It comes from McKinsey’s article “New front door to the internet: Winning in the age of AI search,” published October 16, 2025. The article states that $750 billion in US revenue or consumer spend will flow through AI-powered search by 2028 and footnotes it as a McKinsey projection. The published article doesn’t explain how the projection was built or which spending it includes.

Is the $750 billion projection reliable?

It is a named source’s estimate without a published method, so you can’t check it. The same page describes it as revenue AI search could impact and as spend that will flow through AI search, which are different claims. Treat it as a sign that a large consultancy expects AI search to touch a large share of buying, not as a figure to plan budgets or forecasts against.

How many consumers use AI search?

In McKinsey’s August 2025 survey of 1,927 US consumers, half said they intentionally seek out AI-powered search, and 44 percent of AI search users called it their primary and preferred source of insight, ahead of traditional search at 31 percent. The numbers are self-reported and US-only, but the sample and timing are published, which makes them more usable than the revenue projection.

How much search traffic could brands lose to AI search?

McKinsey’s analysis puts the possible decline in traditional search traffic for unprepared brands at 20 to 50 percent, without showing how the range was calculated. A spread that wide is best read as a scenario rather than a forecast. Your own data is a better guide: compare organic clicks and impressions over time for queries where AI Overviews appear against those where they don’t.