Search did not stop working in 2026. It stopped sending traffic.
SparkToro’s analysis of Similarweb clickstream data found that 68.01% of US Google searches in the first four months of 2026 ended without a click, up from 60.45% in 2024 (UK, German, French, Italian and Canadian figures are still to come, but the direction is not in dispute). That is a 7.56-point jump in two years — the fastest acceleration in a decade. Clicks to the open web fell 9.51 percentage points over the same window, roughly a 23% reduction in the traffic reaching independent websites. AI Overviews now appear on more than 20% of searches, and where they appear, click-through rates fall by close to 60%.
The obvious conclusion — that SEO is finished — is the wrong one, and the data that disproves it is the most important number in this article.
The 1% that Behaves Like 10%
Conductor’s 2026 AEO/GEO benchmark study, across ten industries, found that AI answer engines currently drive 1.08% of total website visits. ChatGPT accounts for 87.4% of that referral traffic. Growth is running at roughly 1% month on month. In IT specifically the figure is 2.80%; in consumer staples, 1.91%.
One per cent sounds like a rounding error. It is not, because of what that one per cent does after it arrives. Referencing audience journey data from Knotch, the same study reports that visitors referred directly from LLMs convert at twice the rate, in one-third the number of sessions.
That is the whole strategic case for answer engine optimisation in a sentence. A person who arrives from an AI assistant has already had the comparison conversation. They have been told what to look for, they have been given a shortlist, and they are arriving to verify a decision rather than to begin researching one. They are, functionally, a referral from an advisor rather than a click from a results page.
If your brand is not in that conversation, you are not losing 1% of traffic. You are losing the pre-qualified end of your pipeline, and you are losing it invisibly, because nothing in Google Analytics reports the deals you were never shortlisted for.
What Answer Engine Optimisation Actually Is
Answer engine optimisation (AEO) — also called generative engine optimisation (GEO) — is the practice of making your content the source an AI system quotes when it answers a question in your category.
It shares infrastructure with SEO and diverges in objective. Classical SEO optimises for a position on a page of ten blue links. AEO optimises for inclusion in a synthesised answer, where there is no page, no position, and frequently no click. The unit of success is a citation.
Three practical consequences follow.
- Ranking still matters, but as an input rather than an outcome. Answer engines draw heavily on well-ranked, crawlable pages. Technical health, indexation and topical depth remain the entry ticket. What they no longer guarantee is a visit.
- Format determines extractability. Conductor’s data on the most-cited page types in AI Overviews puts blog content first, followed by video, articles, news and product pages. What these have in common when they get cited is structure: a question stated plainly, an answer given immediately beneath it, and supporting evidence attached. A 2,000-word essay that buries its conclusion in paragraph fourteen is a poor citation candidate regardless of quality.
- Entity clarity beats keyword density. An answer engine must be confident about who you are, what you do and where you operate before it will recommend you. That confidence is built from consistent, corroborated signals across your own site and across third-party sources it trusts — which is why off-page work and digital PR have quietly become AEO work.
The AEO Playbook
1. Measure Your Baseline Visibility, Not Just Your Rankings
Track how often your brand is named in AI answers for the twenty or thirty questions that precede a purchase in your category. Run those prompts across ChatGPT, Gemini, Perplexity and Claude on a fixed schedule and log whether you appear, whether a competitor appears, and which source was cited. Without this baseline you cannot prove the work is doing anything.
2. Build Question-First Content Architecture
Take the questions your sales team answers weekly and give each one a section — H2 as the question, a direct 40–60 word answer immediately below it, then the detail, examples and data. This is the single highest-return structural change most sites can make, and it improves human readability at the same time.
3. Attach Evidence to Every Claim
Answer engines preferentially cite content that carries verifiable specifics — named studies, dated figures, original data. “Conversion rates improve” is unciteable. “AI-referred visitors converted at twice the rate across ten industries (Conductor, 2026)” is a sentence a model can safely repeat. Original research, even small-scale, is the most defensible asset you can build here.
4. Implement Structured Data Properly
Organisation, Article, FAQPage, Product and LocalBusiness schema give machines an unambiguous reading of your pages. Schema is not a ranking lever; it is a comprehension aid, and comprehension is what determines whether you are eligible to be quoted.
5. Earn Corroboration Off-Site
Listings, credible directories, industry publications, expert commentary and community answers all feed the models’ picture of your brand. A brand described consistently across many independent sources is a brand an answer engine can recommend without hedging.
6. Rebuild Your Reporting Around Assisted Value
If your dashboard rewards sessions, AEO will look like a failure for two quarters. Report on citation share, branded search volume, and the conversion rate and velocity of AI-referred sessions relative to other channels. That last comparison is where the case for continued investment lives.
The UK, UAE and India Picture
The mechanics are global. The commercial reality is not.
United Kingdom. The UK market combines high AI-assistant adoption with expensive paid search, which sharpens the arbitrage: every pre-qualified visitor arriving through an answer engine is one you did not pay a competitive CPC for. Regulatory scrutiny of AI answers is also more active here, so accuracy and clear sourcing carry additional weight for regulated categories — financial services above all. Practical implication: prioritise original UK-specific data and named expert authorship, and make your E-E-A-T signals explicit rather than implied.
United Arab Emirates. Dubai’s market is unusually competitive relative to its search volume, and buying decisions frequently involve a shortlist assembled by someone comparing providers cold. That is precisely the moment an AI assistant is now consulted. Two priorities follow: bilingual coverage, since Arabic-language answers draw from a thinner source pool and are therefore easier to influence, and rigorous local entity signals — a complete Google Business Profile, consistent NAP data, presence in credible UAE directories, and content that names the emirate and the free zone context explicitly.
India. India offers the largest volume and the widest gap between search demand and citable content quality. Vernacular and voice queries make well-structured question-and-answer content disproportionately effective, and the cost of producing genuine depth is lower here than in either of the other two markets — which makes India the sensible place to build the original research assets that then earn citations in all three.
Where to Start if You Are Starting Today
Pick the twenty questions that appear immediately before a purchase in your category. Check which of them your brand currently appears in across the major assistants. Restructure the five pages closest to those questions into question-first format with evidence attached, add the correct schema, and re-check in thirty days.
That is a fortnight of work and it will tell you more about your position in AI search than another quarter of ranking reports.
The clicks are not coming back. The buyers are still out there — they are just being introduced to someone before they ever open a browser tab. The only question that matters is whether the introduction mentions you.
FAQ
Publish on-page — these are the FAQPage schema entries.
What is Answer Engine Optimisation?
Answer engine optimisation is the practice of structuring content and brand signals so that AI systems such as ChatGPT, Google’s AI Overviews, Gemini and Perplexity cite your business when answering questions in your category. Success is measured in citations and assisted conversions rather than in ranking positions.
Is AEO Different from SEO?
It shares foundations with SEO — crawlability, indexation, topical authority — but optimises for inclusion in a generated answer rather than for a position in a list of results. In practice AEO adds question-first content structure, evidence-backed claims, structured data and stronger off-site corroboration.
Is AI Search Traffic Worth Pursuing if It Is Only About 1% of Visits?
Yes, on quality rather than volume. Conductor’s 2026 benchmarks put AI referrals at 1.08% of total visits, growing roughly 1% month on month, while LLM-referred visitors convert at twice the rate in one-third the sessions.
How Do I Know Whether AI Assistants Currently Recommend My Brand?
Run the twenty to thirty purchase-intent questions in your category across the major assistants on a fixed schedule and record whether you are named and which sources are cited. That log is your baseline and your reporting line.
How Long Does AEO Take to Show Results?
Structural changes to existing well-indexed pages can affect citations within weeks. Building entity strength and off-site corroboration is a two-to-three quarter programme.
