Every couple of years someone announces the death of SEO, and every couple of years UK businesses carry on winning customers from search. The honest answer in 2026 is more useful than a straight yes or no.
Why people are asking if SEO is dead again in 2026
The question has teeth this time, and it is worth understanding why. When we pulled the UK search results for "is seo dead" in September 2026, Google answered the question itself in an AI Overview before a single organic listing appeared, and the first genuine result underneath it was a Reddit thread. That is the entire argument in one screenshot: the search engine now resolves the query on its own page, and when it does send a click onward it often sends it to a forum rather than a business. Add ChatGPT, Gemini and Perplexity answering the same sort of question in a chat window, and it is reasonable for a business owner to ask whether the whole channel still pays. But "fewer clicks on some queries" is a very different claim from "search no longer sends customers", and conflating the two is where most of the panic comes from. People still search before they buy. What has changed is which searches produce a visit, and what a page has to do to earn one.
What the data actually says about AI Overviews and clicks
Most of the "SEO is dead" commentary runs on vibes. There is real measurement available, and it is worth looking at properly because it settles part of the argument and complicates the rest. The most careful public study is Pew Research Center's analysis of real browsing behaviour, which tracked 68,879 Google searches made by 900 adults. Three findings matter for anyone deciding whether to keep investing.
- AI Overviews roughly halve the click rate, they do not eliminate it. Users who saw an AI summary clicked a traditional result on 8% of visits. Users who did not see one clicked on 15%. That is a serious cut, and it is also a long way from zero.
- Being cited inside the summary is not a traffic strategy on its own. Only 1% of visits to a page with an AI summary produced a click on a link inside that summary. Citation buys you visibility and credibility. It does not reliably buy you a session.
- They appear on a minority of searches, and a predictable minority. In that dataset 18% of searches produced an AI summary. It rose to 53% for searches of ten words or more and 60% for searches starting with a question word, but only 8% of one and two-word searches triggered one.
That last point is the one almost nobody quotes, and it is the most commercially useful. Long, conversational, question-shaped queries are the ones getting summarised. Short, blunt, high-intent queries are largely not. "How does search engine optimisation work" is in the blast radius. "Seo agency taunton" is not. Two caveats before you build a plan on these figures: the panel was US-based, so UK behaviour may differ at the margins, and the fieldwork ran in March 2025, since when AI Overview coverage has broadened. Treat the direction as solid and the precise percentages as dated.
Six things about SEO that genuinely did die
The people declaring SEO dead are usually right about something specific. These six practices really are finished, and if your strategy still leans on them, your results will look like a dying channel.
Writing for keyword density
Repeating a phrase a target number of times stopped working years ago and now actively hurts. Modern ranking systems evaluate whether a page answers the question, not whether it contains the string. Pages written to hit a density percentage read badly to humans and add nothing an AI summary cannot already produce.
Publishing thin top-of-funnel volume
The old model of shipping forty short posts a month on broad definitional questions is the single biggest casualty. Those are precisely the long, question-shaped queries that trigger a summary six times out of ten. If a 600-word article tells a reader nothing they could not get from the results page, it will not earn the click.
Buying links at scale
Cheap bulk link packages were already a liability. They are now a fast route to a manual action with no upside, because authority increasingly comes from being genuinely referenced by sources that people and models already trust.
Ranking without a brand
An anonymous site with no reviews, no press, no named authors and no real-world footprint has very little to hold on to in 2026. Both search engines and language models lean on corroboration from independent sources, and a brand nobody mentions gives them nothing to corroborate.
Treating sessions as the only scoreboard
Judging SEO purely on total organic sessions makes a healthy channel look like a failing one. If informational traffic falls while enquiries hold steady, that is a mix shift, not a decline. Reporting that cannot tell those apart will keep producing false alarms.
Set-and-forget technical SEO
A technical audit done once at launch and never revisited is worth very little. Crawl behaviour, rendering, structured data and page performance all drift as a site changes, and the gap only shows up when something has already broken.
What did not die, and shows no sign of it
Strip away the practices above and a large, healthy part of the channel is left standing. Commercial searches still convert, and they are the ones that pay the bills. Somebody typing "emergency plumber Bristol" or "shopify developer uk" or "best CRM for small business" is close to spending money, and those results pages still send clicks because the searcher needs to compare real options, check real prices and contact a real company. An AI summary can describe what a plumber does. It cannot be your plumber.
Notice too that these are short queries. On the Pew numbers, a two or three-word commercial search is one of the least likely things to trigger a summary at all. The queries that lost the most clicks and the queries that produce the most revenue are, for most SMBs, almost entirely different sets.
Three areas are, if anything, stronger than they were:
- Bottom-of-funnel and comparison pages. Alternatives, versus pages, pricing pages, service pages and case studies all serve evaluation intent, which resists summarisation because the searcher wants to make a choice, not read a definition.
- Genuinely original material. First-hand data, real client outcomes, tested processes and specific numbers give both readers and models something they cannot generate. It is also what gets cited. Our breakdown of the content types that win in AI search goes through which formats survive this and which do not.
- Technical foundations. Crawlability, clean structure, fast pages and accurate structured data are the price of entry for both classic results and AI answers. If a model cannot parse your page, it cannot quote you.
This is also why the underlying case for the channel has not changed much. If you want that argument laid out properly, we covered it in why SEO is important for business.
Is local SEO dead too?
No, and it is the part of the channel least touched by any of this. A local search resolves to a map pack, and a map pack is not something a language model can replace, because the searcher needs a real address, real opening hours, real reviews and a phone number that a person answers. Google has strong commercial reasons to keep that surface intact, and the queries behind it are exactly the short, high-intent kind that rarely trigger a summary.
What has changed locally is the standard. A thin, unclaimed listing with four reviews and no photos loses to a maintained one far more decisively than it did five years ago, because there is less room on the page and more competition for it. If anything, local is where the gap between businesses that do the basics and businesses that do not has widened. Our local SEO guide for small businesses covers the work that actually moves it.
Search stopped meaning Google alone
The other half of the shift gets less attention than AI Overviews but matters more over a five-year horizon. People now start product and service research in several places: ChatGPT, YouTube, TikTok, Reddit, Amazon, LinkedIn and comparison sites, as well as Google. Pew's data captures one corner of this indirectly, in that Wikipedia, YouTube and Reddit together accounted for 15% of the sources cited in the AI summaries studied. The engines are quoting community and reference platforms heavily, which tells you where they think the trustworthy material lives.
The practical consequence is not that you need to be on every platform. It is that your brand needs to exist in places other than your own website, because that is what both people and models check before believing you. Reviews, directory entries, mentions in trade press, an active profile where your customers actually are. None of that is new advice. It is simply now load-bearing rather than optional, and it is why "ranking without a brand" is on the dead list above.
How AI Overviews actually change the maths
The useful way to think about AI search is that it splits your old traffic into two piles. Queries that were only ever seeking a fact get answered on the results page and largely stop sending visits. Queries where the searcher has to evaluate, trust or buy something still send visits, and often better-qualified ones, because the tyre-kickers have been filtered out earlier. Google documents how its AI features surface and link to web content, and the mechanics matter: pages still get cited, and citation is now a visibility asset in its own right even when the click does not follow immediately.
That creates a second job alongside ranking, which is making sure your brand is the one an answer engine reaches for. We break down how that works in our guide to getting your brand cited by LLMs, the practical implementation in the AI search optimisation playbook, and where the two disciplines genuinely diverge in AEO vs SEO. The important point is that this is an addition to SEO, not a replacement for it. The technical and content work that gets you cited by a model is very largely the same work that got you ranked.
A five-minute test: is your SEO dead, or just badly aimed?
Before concluding the channel is finished, check whether the problem is the channel or the strategy. Run through these in your own Search Console and analytics:
- Split your queries by intent. Separate informational queries from commercial ones. If informational impressions are down but commercial clicks are flat or rising, SEO is working and your content mix is simply out of date.
- Sort your queries by word count. This is the quickest proxy for AI Overview exposure. If your losses are concentrated in long, question-shaped queries and your short commercial terms are steady, you are seeing the summary effect, not a ranking problem.
- Check position, not just traffic. If your average position is stable and traffic has fallen, you are losing clicks to on-page answers. If position has fallen, that is a ranking problem with a different fix.
- Look at what actually ranks. Count how many of your top pages are thin definitional posts. If most of your organic footprint is that, the decline is a content-strategy problem you caused.
- Check whether enquiries moved. Compare organic enquiries or sales, not sessions, year on year. This is the only number that settles the argument.
- Check your attribution. A growing share of AI-influenced visits land as direct or unattributed traffic, which quietly understates search. Our guide to tracking AI traffic in GA4 covers what standard reporting still misses.
In most audits we run, points one and four explain the whole story. The site is not being punished by AI, it is over-invested in exactly the content type AI made redundant.
What to actually do about it in 2026
The strategic response is a reallocation, not a retreat. Concretely:
- Move budget down the funnel. Fewer definitional posts, more service pages, comparison pages, pricing transparency and case studies. These convert better and are far harder to summarise away.
- Consolidate rather than publish. Merge overlapping thin posts into one substantial page. Two mediocre articles competing for the same query is a self-inflicted wound, and it is extremely common on sites that published to a monthly quota.
- Add something only you have. Your own data, your own results, named experts, photographs of real work. This is the difference between a page a model quotes and a page it ignores.
- Build the brand off your own site. Reviews, trade press, directory accuracy, a presence where your customers already talk. This is what gives an answer engine something to corroborate.
- Fix the technical basics and keep them fixed. Google's own SEO starter guide still describes the fundamentals accurately, and they have not changed as much as the discourse suggests.
- Change what you report on. Track qualified enquiries, assisted conversions and citation visibility alongside rankings. If your reporting only counts sessions, it will keep telling you the channel is dying.
Will SEO still exist in five or ten years?
The label might not. The job almost certainly will. Strip the acronym away and SEO is the work of making a business findable, parseable and credible to whatever system people use to decide who to buy from. That system was a directory, then a link graph, then a ranking algorithm, and it is now partly a language model. Every one of those transitions was announced as the death of the discipline, and each time the underlying job survived because the commercial need behind it never went anywhere.
What is genuinely uncertain is the traffic model. If answer engines keep absorbing informational demand, a website's value shifts further from "receives visits" towards "is the source that gets quoted and the destination people arrive at ready to buy". That is a real change and it will hurt publishers who monetise volume far more than it hurts service businesses who monetise enquiries. For a UK SMB, the sensible planning assumption is fewer, better-qualified organic visits rather than a cliff, and a growing share of influence you can see the effect of but cannot fully measure. Plan for measurement getting harder, not for demand disappearing.
What the "SEO is dead" argument gets wrong about budget
The commercial version of this question is not really "is SEO dead", it is "should I move this money to ads". Worth being clear about the trade: paid search buys visibility that stops the moment you stop paying, and its cost per click has not got cheaper. SEO buys an asset that keeps returning after the invoice is settled, at the cost of a slower start and no guarantees. Neither is a complete strategy on its own, and the businesses doing best in UK search right now generally run both, using paid to cover the terms they cannot yet win organically.
What has changed is the sensible payback expectation. Six months to meaningful movement is realistic for most SMBs, longer in genuinely competitive sectors. If you are weighing up whether to spend at all, we have set out the honest version in our guide to SEO ROI and the numbers in our UK SEO pricing guide. If you would rather have someone look at your specific situation, that is what our SEO services are for.
Frequently asked questions
Is SEO dead in 2026?
No. Search still sends commercial traffic, and businesses ranking for buying-intent queries continue to get enquiries from it. Pew Research Center found that users who saw an AI summary still clicked a traditional result on 8% of visits, against 15% for users who saw no summary. That is a meaningful cut, not a collapse. What has died is the specific version of SEO built on thin keyword-led content, bought links and traffic-only reporting. If that describes your strategy, your SEO is dead. The channel is not.
Have AI Overviews killed organic traffic?
They have roughly halved the click rate on the searches where they appear, and they appear on a minority of searches. In Pew's dataset 18% of Google searches produced an AI summary, rising to 53% for searches of ten words or more but only 8% for one and two-word searches. Because long, question-shaped queries are the most exposed and short commercial queries the least, the effect lands hardest on informational content and lightest on the pages that generate enquiries. Sites whose organic footprint was almost entirely informational have taken a genuine hit.
Is local SEO dead?
No, and it is the least affected part of the channel. Local searches resolve to a map pack, which needs a real address, real opening hours and real reviews, none of which an answer engine can substitute for. Those queries are also typically short and high-intent, so they rarely trigger an AI summary. What has changed is the standard: a thin, unclaimed listing now loses far more decisively to a well-maintained one than it did five years ago.
Will SEO still exist in five years?
The label may not survive, but the job will. SEO is the work of making a business findable, parseable and credible to whatever system people use to choose a supplier, and that system has already been a directory, a link graph and a ranking algorithm before becoming partly a language model. What is genuinely uncertain is the traffic model. Plan for fewer but better-qualified organic visits, and for measurement getting harder, rather than for demand disappearing.
What is replacing SEO?
Nothing is replacing it, but something is being added to it. Generative engine optimisation, sometimes called answer engine optimisation, is the work of getting cited inside AI answers. It overlaps heavily with technical SEO and content quality, so it is better understood as an extension of the discipline than a successor to it. Be realistic about what citation buys you: Pew found that only 1% of visits to a page with an AI summary produced a click on a link inside that summary, so citation is a credibility asset more than a traffic one.
Is SEO still worth the money for a small business?
For most UK SMBs, yes, provided the work is aimed at commercial queries rather than blog volume. The economics are unchanged: a page ranking for a buying-intent term keeps producing enquiries with no incremental media cost, and those short commercial terms are the ones least disrupted by AI summaries. The mistake is buying a content-quota package and expecting it to behave like a lead-generation programme.
