Half of us now use AI search (and half of ‘traditional search traffic’ is at risk)

45% are using those tools for purchasing decisions, while 44% say AI search is their “primary and preferred source of insight” – well ahead of traditional search at 31%.

That shift is across age demographics, with “a majority of older generations,” including baby boomers, using the tools (although boomers are the only demographic that still prefer search engines to AI search).

As this trend only continues to deepen, the research projects that $750bn of consumer spend will “flow through” AI-powered search by 2028 – but brands, it says, are not ready for this shift. Citing a “readiness gap,” the research finds that between 20% and 50% of even top brands’ traffic from traditional search channels is at risk, while only 16% of brands are systematically tracking their performance in AI search.

The new search landscape

For a long time, ‘search’ was almost coextensive with a single entity – Google – allowing search marketing practitioners to go far with deep expertise in that single platform.

But the landscape is becoming far more heterodox.

Even within ‘traditional’ search, things are changing. Google’s dominant 93% market share from 2023 fell to below 90% this year as the giant ramps up AI integrations into its standard AI products (per McKinsey analysis, 50% of Google searches already serve AI summaries – a number that will rise to 75% by 2028).

Meanwhile, the phenomenon of ‘social search’ is going stratospheric, too – especially in the brand space. 75% of us have now used social media to search for new products and brands, while the likes of TikTok are closing the loop on efforts to make their platforms one-stop shops for marketing and sales alike.

• How AI search is remaking the B2B world, too

It’s against this backdrop that generative model providers are making a play for an even greater share of this newly expanded search pie. OpenAI’s ChatGPT now counts 700 million weekly users. With its eyes set on a billion, it recently launched its first global brand campaign, which showcases people using the software for the purposes they may have once used traditional search engines for: finding recipes and work-out plans (it came just a week after Anthropic’s first major campaign for Claude).

Practitioners in the search world haven’t been slow to spring up offerings of ‘generative engine optimization’ (GEO), promising to help brands make sense of this brave new world. The McKinsey release praises this shift, arguing that brands must invest in GEO as a “core capability” but insists that with AI search “rewriting the rules of visibility and marketing across the consumer decision journey,” there’s a task ahead to “adjust to this new reality.”

The new world of GEO

According to the McKinsey report, the nascent world of GEO demands that brands “rethink their approach to both digital content and organic search.”

The core challenge is understanding the variety of different large language models and the roles of different content types in influencing and showing up in the outputs of those models. Brands’ own channels only make up 5-10% of sources that AI search results draw from, the research finds – a fact that demands a rethinking of content strategy.

While SEO focuses on owned channels, in other words, LLMs pull from large and evolving pools of sources, which “makes meeting consumers in the places where they make decisions more difficult.” The upshot is that even major brands and market leaders don’t show up in AI-powered search results.

The challenge for brands, then, is to understand as best they can the varied models of the dominant generative models, keep paying attention as they evolve and shift strategy to appear throughout the wide web of sources that those models draw upon.

“To influence the myriad sources utilized by AI-powered search and inform the marketing road map, brands need to understand what questions consumers are asking and which sources are shaping answers,” the McKinsey report says. “And they need to continually keep pace as LLMs evolve.”