How AI is changing, and might kill, SEO

Welcome to Artificial Intelligence (AI), the new front door of discoverability to the internet. One that may be slamming shut on the old architecture of search engine optimisation (SEO).

McKinsey's latest analysis reveals that AI-powered search is no longer a novelty but the default pathway for consumer discovery. Half of Google searches now feature AI summaries, a figure poised to surpass 75 per cent by 2028. Consumers across demographics, including a surprising majority of baby boomers, are flocking to tools like ChatGPT, Perplexity and Copilot. 44% deem AI as the premier source for purchase decisions.

By 2028, McKinsey forecasts $750bn in US revenue will course through these channels. For media buyers and brands, the implication is stark: the era of "zero-click" searches is upon us, eroding traffic by 20-50 per cent and forcing a radical rethink of search engine optimisation (SEO).

This seismic shift is not hypothetical. The 2025 Previsible AI Traffic Report lays bare the numbers: referrals from AI platforms surged 527 per cent in the first five months of the year. In consultative sectors such as finance and health, these now comprise up to 1% of sessions, a foothold once reserved for Google alone.

Publishers have fared worst, with click-through rates plummeting by more than 50% in some cases. At the heart of this disruption lies a profound behavioural pivot. Users are abandoning keyword hunts for conversational probes, blending voice, vision and context into queries that AI resolves on the spot.

Over 70 per cent of AI queries occur at the top of the funnel, McKinsey reports, yet they span the entire journey: from category exploration to feature comparisons and personalised recommendations. In consumer electronics, 55% of shoppers rely on AI search; in apparel, it's 40%. The result? Traditional SERPs (search engine results pages) are bypassed, rendering the blue-link economy obsolete.

SEO, long the linchpin of digital visibility, is in freefall. Where once marketers chased page-one dominance, they now vie to be cited, or even mentioned, in synthesised summaries.

Brand-owned sites contribute a mere 5-10% of AI's source material, McKinsey notes, with the rest harvested from affiliates, forums and user-generated content. Even market leaders vanish from responses; top credit-card issuers, for instance, are routinely absent despite their dominance.

Adaptation is underway, though uneven. Only 16% of brands systematically track AI performance, McKinsey finds, leaving most exposed. Savvy players are prioritising "Gen AI engine optimisation" (GEO), a complement to SEO that audits visibility across LLMs and sentiment in outputs.

To make way for the shift, content strategies must broaden: invest in third-party channels like TikTok and Instagram, where younger cohorts source 40% of insights, and forums that feed AI's "answer graph". Here, Wikipedia and Reddit play an important role. As do media like AP, Reuters, BBC, NPR, CNN, NYT, and Washington Post.

Quality trumps quantity. Human-centric depth, featuring unique expertise, structured data, and conversational phrasing, boosts citability.

My advice to every CMO reading this is uncomfortably simple: treat your website as a source, not a destination. Accept that most prospects will meet your brand first as a single sentence inside someone else’s synthesis. Make that sentence indisputably the best one available. Earn the citation, not the click. And start budgeting for a world where the media plan must buy attention inside the answer, not beside it.

Search marketing is not evolving. It is being replaced. The only question left is whether we have the courage to bury the corpse and build something new on the grave.