Organic clicks are down. I can prove that part. Almost everything else being said about it is unprovable, including by the people saying it loudest.
Why can’t anyone agree?
Because the loudest numbers in this debate are the worst ones. The most-quoted prediction of the last two years – that search volume would fall 25% by 2026 – was scenario modelling by an analyst firm, published as a projection, laundered into a fact by a thousand blog posts, and it did not happen. Meanwhile Google says clicks are “relatively stable” and of higher quality, and has never published the data behind it.
So both poles of the argument are being argued from sources that would not survive a second look. That is worth knowing before you move a budget on the strength of a statistic in a LinkedIn post.
What has actually changed?
Two things, and they are well evidenced. Zero-click search is real and rising: independent clickstream analysis put 68% of US Google searches ending without a click in early 2026, against roughly 60% in 2024 – about 276 clicks per thousand searches, down from 374. The people who published it flag that their panel providers changed between those years, which is the kind of caveat you should look for and rarely get.
And AI summaries are associated with far fewer clicks. Pew watched what 900 people did across 68,879 searches: with an AI summary on the page, a result was clicked on 8% of visits, against 15% without one. Only 1% clicked a link inside the summary itself. It is the best-evidenced finding in this argument, though not a clean experiment – summaries show up more on longer, question-shaped queries, which are not the same queries.
What has not changed: people still search, in enormous volume, and the searches that were closest to a transaction are still the ones most likely to produce a click. The volume of organic clicks is falling. The value of the ones that remain is probably rising. Nobody can prove the second half yet, and you should be wary of anyone who claims to.
Which parts of the old playbook are dead?
Publishing to capture informational keyword volume. If the question can be answered in a paragraph, an answer engine will answer it in a paragraph, and your article about it will be read by a model rather than a buyer. Ranking reports as the deliverable, because a position that produces no click is a vanity metric with a chart. Content velocity as a strategy. And treating organic as a separate line item with its own agenda, disconnected from what you are paying for on paid.
“A ranking that produces no click is a vanity metric with a chart attached.”
What still compounds?
Branded and navigational demand, which is the traffic nobody can take from you. Local and near-me intent. Transactional and comparison queries, which still generate clicks because a summary cannot complete a purchase. And genuine first-party depth – prices, availability, real constraints, the specifics of what you actually sell – which is the one category of content a model cannot invent and a competitor cannot copy without lying.
That last category is also what gets you cited in the answer layer. Same fixes, twice the return – which is the whole of our argument about what AEO actually is.
Where does SEO earn its place next to paid?
The cleanest argument I know is displacement: what would you have paid Google for the same demand? If a query you rank for costs £4 a click on paid and delivers a hundred sessions a month organically, that is a £400 monthly line you are not paying – and unlike the ad, it does not stop the day you pause it.
The second argument is that the two channels teach each other, and the exchange only happens if somebody is holding both. Run them in separate silos with separate reports and you throw it away – which is a large part of why we run every surface as one connected system rather than as a menu of retainers.
Is it still worth it if you’re small?
Sometimes, and not always, and the honest answer depends on which queries you are trying to win. If your ambition is to out-publish an established competitor on broad informational terms, no – you will spend a year losing to somebody with a decade of authority and a bigger team, and an answer engine will summarise the topic before either of you gets a click.
Where a smaller operator genuinely wins is the narrow end. Your own brand terms, which nobody should be beating you on and which are cheap to defend. Local and near-me intent, where relevance beats domain strength. And the specific, awkward, long-tail questions about what you actually sell – the ones with real answers that only you have. That is a much smaller surface than the SEO industry would like to sell you, and it is defensible in a way the broad end no longer is.
Budget for quarters, not weeks. If you need a result inside a month, that is a paid search question, and anyone telling you otherwise is selling a retainer.
How do you measure it when the click isn’t the unit?
You stop reporting rankings and start reporting substitutes, in roughly this order of rigour. Geo holdout tests on organic-heavy segments, if you have the volume for them. Branded search and direct sessions, which is where an answer you never saw tends to show up a fortnight later. Paid CPC displacement, as above. And then, weakest but not worthless, citation tracking – a fixed prompt set run monthly across the main assistants, logging whether you are named and whether the description is accurate.
Know the ceiling on the platform data. Search Console’s generative AI report, added in June 2026, gives impressions only – no clicks, no click-through rate, no queries.
One more thing, because it cuts against the hype in both directions. When Amsive looked at six months of analytics across 54 sites, traffic arriving from LLMs converted at 4.87% against organic’s 4.60% – slightly better, but nowhere near statistically significant, and under 1% of sessions. So the vendors claiming assistant traffic converts several times better are overreading noise, and so is anyone telling you it converts worse. It is worth having. It is not yet worth restructuring around.
What does this look like in practice?
Fewer pages, better ones. Depth where you have genuine first-party knowledge and silence where you do not. Organic measured on the demand it captures rather than the positions it holds. And a working relationship between the organic and paid sides of the account, so that what one learns the other uses.
It also means each remaining session has to work harder, which is the argument for spending less time on rankings and more on what happens after the click. Fewer visitors making the same number of bookings is a perfectly good year.