Is SEO Dead? What the 2025–2026 Click Data Actually Says

"Is SEO dead" has been asked annually for 20 years, and the honest answer has always been no. This time the question deserves more respect: the 2025 click data shows a real, measured structural change, not a panic cycle. Search did not die — but a specific business model built on it is in genuine trouble, and pretending otherwise would make everything else on this site less credible.

We run 3 production sites built for this transition — answer-first from their first commits — so what follows is the measured public record first, and how we operate inside it second.

What does the click data actually show?

It shows clicks per search falling wherever AI answers appear, in every study that measured it:

Sources: Ahrefs, Seer Interactive, Pew Research Center.

Those are different methodologies agreeing on direction, which is the strongest signal this niche produces. The full curated set — including how often AI summaries appear at all — lives on our AI search market share data page.

Is anything recovering?

Yes — two findings complicate the obituary. Seer Interactive's 2026 update documents a partial CTR rebound in early 2026, softening the steepest 2025 declines (Seer, 2026). The system is still moving; December 2025 was not the end state.

Second, and more strategically useful: in Seer's 2025 data, brands cited inside an AI Overview saw a +35% higher click-through rate than uncited brands on the same queries (Seer). The click penalty is not evenly distributed — it concentrates on pages that are neither the answer nor named in it. Visibility did not disappear; it moved into the answer box and got harder to share.

If clicks fall, what replaces them?

The citation — being the source the answer engine quotes and links — becomes the second unit of visibility alongside the ranking. That is not a soft-focus consolation prize; it is a measurable position with measurable behavior attached: a +35% CTR lift when you hold it (Seer, 2025), and a hard ceiling on volume, since Pew found just 1% of AI-summary encounters produced a click on a cited source (March 2025).

Whether that trade is worth pursuing depends on your economics, and we argue both sides in are AI citations worth anything. What it is not is a separate discipline sold under a new acronym — the work that earns rankings and the work that earns citations is substantially the same page.

What still works exactly the same?

The foundation, according to the platform that runs it. Google's guidance on succeeding in AI search says to apply the same foundational SEO best practices for AI features as for Search overall, and its AI-features documentation is explicit: "There's also no special schema.org structured data that you need to add," no AI text files, no new markup — eligibility is being indexed and snippet-eligible (Google, May 2025; AI Features and Your Website).

So crawlability, indexing, site structure, entity clarity, and pages that answer real questions did not depreciate — they are the eligibility layer for both surfaces. A team that was doing that work well has lost no asset. A team that was mass-producing thin informational pages to harvest clicks has lost the arbitrage, and no relabeling of SEO to GEO restores it.

What did we reallocate on our own sites?

Three commitments, stated as operations rather than philosophy [our data]:

• We build pages to be liftable. Every page in our fleet leads with a direct answer sized for extraction, on the logic that if the answer box is taking the impression, we want to be its source. Our documented AI Overview citation win — a new glossary page built this way, cited within days of shipping — is the pattern's receipt.
• We built measurement for the new surface. GSC cannot show AI Overview data and GA4 cannot see citations, so we run our own instrumentation — a GA4 AI-referral channel and Search Console query mining — on our own builds.
• We do not write for queries whose click is the only value. Where an AI answer fully satisfies the query and our page would be neither cited nor clicked, we do not compete. That is a real reduction in addressable queries, and we treat it as one.

What we did not do: abandon publishing, block the AI crawlers in a fit of principle (that decision deserves an actual framework — ours is here), or buy the panic. Search demand did not shrink; its payout structure changed.

Whose SEO actually is dead?

Against our own interest as people who sell content builds: if your model is high-volume informational pageviews monetized by display ads, the measured CTR data is a structural problem no optimization fixes. A −34.5% to −58% click decline on informational queries (Ahrefs, March–December 2025) is not a tactics gap — it is the platform consuming the impression your revenue depended on. The honest options are changing what the traffic is for (products, services, lead capture) or accepting a smaller channel; hiring an SEO or GEO vendor to restore 2023 pageviews is money spent against the direction of the data.

For everyone else — businesses where search feeds a sale rather than an ad impression — SEO is not dead. It is bigger by one scoreboard: the same disciplined page now competes for the ranking and the citation. How to build for both is the whole of the operator's guide to generative engine optimization.