The complaint arrives in a predictable shape. Rankings look the same or better, impressions are flat or up, and clicks are down. Nothing is broken, no penalty landed, no competitor obviously overtook you — the traffic simply thinned out, and every dashboard insists everything is fine.
That gap between visibility and visits is the practical face of a genuine shift: a growing share of searches now get resolved on the results page rather than on a website. The answer is assembled, summarised, or displayed in place, and the click becomes optional.
The takeaway up front: this does not make SEO obsolete, but it does split the job in two. One job is being the source an answer gets built from. The other is being worth a visit anyway. They need different work, they succeed in different ways, and if you keep measuring both with a single traffic number you will misread your own site for a year.
What actually changed, and what only feels new
It helps to separate the mechanism from the drama, because the mechanism is old and stable.
Search engines have been absorbing answers for a long time. Definitions, conversions, opening hours, sports scores, weather, calculations, the spelling of a word — these stopped being clicks years ago, and almost nobody mourned them, because they were never traffic worth having. What has changed is the width of that behaviour. Answers assembled from several sources now cover questions that used to require reading a page: comparisons, how-tos, "what is the difference between", multi-step explanations.
Two things follow directly, and neither is a prediction:
- The threshold moved. The kind of question a results surface can finish on its own used to be narrow and factual. It now includes a good deal of explanatory content — which is exactly what most informational pages are made of.
- Being read and being visited came apart. Your material can inform an answer that never sends anyone to you. That is a real loss of clicks and, separately, a real gain in reach. Both are true at once, and only one shows up in your analytics.
What has not changed is worth stating just as plainly, because overreaction is expensive. Content still has to be found, fetched and understood. Sites still get chosen for their credibility. Clear structure and internal linking still decide which of your pages is treated as the definitive one. The foundations laid out in our SEO fundamentals guide did not become obsolete; they became the entry fee rather than the strategy.
Which pages lose, and which do not
The loss is not evenly spread. It concentrates almost entirely in one category, and knowing which of your pages sit there is the most useful hour of analysis you can do.
Pages whose entire value fits in a sentence lose. If a page exists to answer "what does this term mean", "how many centimetres in an inch", "what is the standard size of X", its value transfers cleanly into a summary and the visit stops being necessary. That is not unfair; it was always a fragile business.
Pages that require the reader to do something do not. Six kinds hold up consistently, and the common thread is that a summary cannot substitute for them:
- Decisions with trade-offs, where the honest answer is "it depends" and the value is in helping someone work out which case they are in.
- Original material — your own data, your own process, your own worked examples. A summary can point at it; it cannot regenerate it.
- Tools and interactive things, where the point is doing, not reading.
- Judgement and stance, including the parts of a subject where experienced practitioners disagree.
- Anything transactional, where the visit is the point.
- Depth, where the answer genuinely takes a while and a paragraph flattens it into something misleading.
Run your own top pages against that split. The exercise usually reveals that a site's traffic decline is concentrated in a handful of definitional posts, while the pages that actually produce enquiries are untouched — which is a much calmer problem than "our traffic is falling".
The two jobs, and what each one asks of you
Job one: be the source
If answers are assembled from sources, then being reliably usable as one matters. This is less mysterious than it sounds, and most of it is old technical hygiene with a new payoff.
- Be reachable. Anything assembling an answer has to fetch the page first. Blocked crawling, aggressive bot rules and broken redirects remove you from consideration silently — the failure modes are covered in what actually blocks a crawler from reaching your pages.
- Answer the question plainly, early, in the page's own words. A page that states its conclusion in two clear sentences near the top is easier to quote correctly than one that builds towards a conclusion for a thousand words.
- Make the structure match the questions. Headings phrased as the things people actually ask, one idea under each, with the answer directly beneath.
- Be specific enough to be worth citing. Generic text that restates common knowledge has no reason to be picked over anyone else's generic text. A concrete detail, a named constraint, a stated exception — those are the parts that survive summarisation with your fingerprints on them.
- Be attributable. A clear, consistent identity — who publishes this, what they do, where else that name appears — is what lets a system treat your material as coming from a known entity rather than an anonymous page.
None of that is a trick. It is the same advice that made pages readable by humans, with a second beneficiary.
Job two: be worth the visit anyway
The second job is the one people skip because it is harder. If a summary can finish the question, the only reason to come to you is that something on your page cannot be summarised.
That leads to a blunt content decision: stop spending your budget on pages that only restate what is already common knowledge. Redirect it towards the things listed above — the decision guides, the original workings, the tools, the depth. In practice this usually means fewer pages, each with more in it, which is also the cheaper way to run a site. The clustering approach in our content strategy guide still applies; what changes is the target for the supporting pages, which now need a reason to exist beyond covering a keyword.
The second half of job two is name recognition. When a results page can answer most things, the searches that reliably reach a site are the ones that name it. Being the answer people go looking for by name is not a hack you can execute this quarter, but every piece of original work, every place your name appears in a context you did not pay for, and every reader who remembers where they read something, contributes to it.
Measure it before you react to it
The fastest way to make a bad decision here is to look at total sessions and panic. Three measurement changes make the picture legible.
Split branded from unbranded. They are now two different businesses. Unbranded informational traffic is the exposed part; branded search is the part that reflects whether people are choosing you deliberately. A site where the first falls and the second rises is in better shape than it looks.
Judge pages by what they produce, not by what they receive. Enquiries, sign-ups, sales, or whatever counts for you, divided by visits. A page that loses casual readers to a summary while keeping the people who were going to act is a page whose conversion rate improves as its traffic falls. That looks like decline in one chart and health in another.
Look at queries, not the sitewide line. Group your queries into the ones a results page can finish and the ones it cannot, and track those separately. This is where you will see the change clearly, and it tells you where to move effort — which the sitewide line never does.
One practical note: take the baseline now. Analysis is only possible against a before, and the split above is much easier to establish while you still have a full picture.
What to do differently on Monday
- Audit for the split. List your top pages, mark each one summarisable or not. That list is your plan.
- Consolidate the thin ones. Several near-identical definitional posts serve you better as one substantial page — fewer, deeper, clearer.
- Add the thing that cannot be summarised. Your own numbers, your own process, a worked example, a tool. One per priority page is plenty.
- Front-load the answer on every page, then earn the rest of the read.
- Fix crawl access if any of it is shaky. Being unreachable removes you from both jobs at once.
- Change the report before you change the content, so you can tell whether anything you do worked.
FAQ
Is SEO still worth investing in if results pages answer the question? Yes, with the effort placed differently. Being findable is now the entry requirement for being the source of an answer as well as for receiving a click, so the technical and structural work matters as much as it ever did. What has weakened is the specific tactic of publishing large numbers of thin pages that restate common knowledge.
Why are my impressions steady while my clicks fall? That is the expected signature of this shift: you are still being shown, and the question is being resolved without a visit. Check whether the decline is concentrated in definitional or quick-fact queries. If your commercial and decision-stage queries are holding, the site is not in trouble — its casual traffic is.
Should I block search engines from using my content? Rarely a good trade for a site that wants customers, because the same access underpins ordinary discovery. Restrictions can make sense for genuinely proprietary material — original data or paid research — but treat it as a business decision about a specific asset, not a sitewide protest.
Does structured data help my content get used in answers? It helps machines read a page accurately, which is a reasonable thing to want regardless of the destination. Treat it as clarity rather than leverage: it describes what your page contains, it does not entitle you to placement.
What kind of content is safest from this? Anything a paragraph cannot replace — decisions with real trade-offs, original data and processes, tools people use, and material where a reader wants a specific source they trust rather than a consensus answer.
The shift is real, and it is smaller than the noise around it. Search absorbed the easy answers; what it did not absorb is judgement, originality and the reason someone chooses one source over another. Sort your pages into those two piles, move your effort towards the second, and change the way you report before you change the way you write. For more practical, no-hype guidance on domains, content, and search visibility, visit MyQSD.
