AI for SEO: The 2026 Playbook

AI for SEO in 2026 means two things at once: using AI to produce and optimise content, and optimising that content to be found by AI. Search no longer runs on one track. The traditional results page with its ranked links still exists, but it is increasingly overlaid with AI-generated answers, and a growing share of searches are resolved inside those answers without a click. Winning now means being visible in both places, and the playbook that gets you there is mostly the old SEO playbook, done with more rigor and a few genuinely new moves.

The mistake to avoid is treating this as either the end of SEO or a brand-new discipline that discards everything. It is neither. The fundamentals that built organic traffic for fifteen years still matter; they are just no longer sufficient on their own. AI Overviews now appear on roughly 48% of Google searches and the rate is climbing (BrightEdge), so the question is not whether AI has changed search, but whether your approach has changed with it.

Search split into two tracks

The structural shift is that search now operates on two parallel tracks, and you have to play both.

The first track is the familiar results page, and it has gotten harder. As AI Overviews absorb more informational queries, click-through rates on the underlying links have fallen: one analysis measured a 58% drop in clicks at the top position across a 300,000-keyword sample (Ahrefs), and a longitudinal study found a 61% decline in organic click-through on informational queries (Seer). For queries that AI can fully answer, the ten blue links are no longer where the attention goes.

The second track is the AI answer itself, and this is where the new game is played. Being cited inside an AI Overview is not the same as being skipped: brands cited in an AI result see about 35% higher organic click-through on those queries overall (Seer), because the citation builds recognition that pays off in later clicks. The goal shifts from ranking first to being the source the AI trusts and quotes. Traditional SEO still wins outright for transactional and tool-based queries, where a user has to click to complete an action, so the honest answer is that you now run both games at once and choose per query which one matters.

What still works

Before the new moves, the reassuring part: most of what worked still works, which is why this is a playbook and not a panic.

Topical authority still wins. Comprehensive coverage of a subject, built as connected clusters with strong internal linking, is if anything more important now, because AI systems favour sources with genuine depth on a topic. Quality and credibility still win; thin, low-effort content was already losing and now disappears entirely. And the foundational rule is unchanged and even sharper: if the content does not exist on your site, you cannot be cited, because AI answers summarise what already exists, and if you did not write it, a competitor who did gets the visibility (HubSpot). The core discipline, publish genuinely useful, authoritative, deep content, did not change. The bar for it rose.

The GEO additions

Generative engine optimisation, or GEO, is the set of new moves layered on top of that foundation, and they follow from how language models read content differently than crawlers do. Crawlers rank pages; language models extract facts, evaluate their credibility, and synthesize an answer. Optimising for the second process means a handful of specific things.

Write for extractability. Make each key point understandable on its own, without the full page for context, so an AI system can lift and quote it cleanly. Clear question-and-answer structure works well, because it maps directly to how queries and answers relate.

Use structured data. Schema markup, Article, FAQ, Breadcrumb, Organisation, helps AI systems understand your content and the entity behind it, and it weighs more heavily in the AI-answer track than it did in classic ranking.

Publish original research. Surveys, benchmarks, and case studies give AI engines something they can only get from you, which is one of the strongest ways to earn citations no competitor can displace.

Build brand mentions and authority. AI systems favour recognised, credible sources, so presence and reputation across the web feed directly into whether you get cited, not just whether you rank.

Keep a refresh cadence. Roughly half of the content cited in AI answers is less than about thirteen weeks old, so freshness is now a ranking-adjacent signal, and GEO is an ongoing discipline rather than a one-time optimisation pass.

Where AI helps you produce it, and the trap

AI is a real accelerator on the production side of this, with one sharp caveat. It is excellent at the research, the keyword and topic-cluster mapping, the competitive gap analysis, and a structured first draft, the volume work that used to make comprehensive content programs expensive.

The trap is using AI to write the finished, published content in its generic voice. The same forces that reward depth and credibility punish thin, obviously-AI-generated text, and the web is now flooded with exactly that, which means generic AI content is a fast way to lose the credibility signals GEO depends on. The winning pattern is the same one that works in cold outreach: AI for the research and the scaffolding, human judgment and expertise for the substance and the voice. AI helps you produce more good content faster; it does not let you skip the part that makes content good.

Own the content and the measurement

The last principle is the one that ties AI for SEO back to every other AI decision: own the asset. Your content, your topical authority, and your measurement are compounding assets you should own, not rent from a platform that can change the rules. That includes measurement, since success is no longer just rankings and clicks but share of model, how often your brand is cited across Google AI Overviews, ChatGPT, and Perplexity for the queries that matter to you. If you cannot see your presence in AI answers, you cannot manage it.

This is one function of the marketing layer inside an owned AI operating system, and it is how content becomes durable demand generation rather than a stream of disposable posts. It is also, candidly, the method behind the library you are reading: connected topic clusters, structured data, cited sources, and a refresh cadence, run as an owned semantic asset rather than a rented campaign. Thane Alaric builds it on the same principle it applies to everything: you own the content, the structure and the measurement, documented and portable, and the models underneath stay swappable. This method is not theoretical here: these Briefings are built with it. If you want the same for your business, tell Daniel what you publish today and what search needs to do for you. Book 30 minutes with Daniel.

Frequently Asked Questions

What does AI for SEO mean in 2026?

It means two things at once: using AI to help produce and optimise content, and optimising that content to be found and cited by AI systems. Search now runs on two tracks, the traditional results page and AI-generated answers, and winning means being visible in both. Most of the old SEO fundamentals still apply; they are now done with more rigor plus a set of new generative-engine-optimisation moves.

Is SEO dead because of AI?

No, but it split into two games. AI Overviews now appear on roughly 48% of Google searches, and click-through on the underlying links has fallen sharply for informational queries. But traditional SEO still wins outright for transactional and tool-based queries, and being cited inside an AI answer drives about 35% higher click-through on those queries. SEO did not die; it now has an AI-answer track alongside the classic results page.

What is GEO, and how is it different from SEO?

GEO, generative engine optimisation, is optimising to be cited in AI-generated answers. It differs from classic SEO because language models extract facts, judge credibility, and synthesize answers rather than ranking whole pages. GEO emphasizes extractable, self-contained points, structured data, original research, brand authority, and freshness. The work overlaps heavily with good SEO; the emphasis shifts towards extractability and citation-worthiness.

How do you optimise content for AI Overviews and ChatGPT?

Write for extractability, so each key point stands alone and can be quoted cleanly, ideally in question-and-answer structure. Add schema markup so AI understands your content and entity. Publish original research AI can only get from you. Build brand mentions and authority, since AI favours recognised sources. And keep a refresh cadence, because roughly half of AI-cited content is under thirteen weeks old.

Can I use AI to write SEO content?

Use AI for research, topic-cluster mapping, competitive analysis, and a structured first draft, the volume work. Do not publish AI’s generic finished writing: thin, obviously AI-generated text loses the credibility signals that both classic SEO and GEO reward, and the web is already saturated with it. The winning pattern is AI for the scaffolding and human expertise for the substance and voice.

How do you measure SEO success in the AI era?

Beyond rankings and clicks, track share of model: how often your brand is cited across Google AI Overviews, ChatGPT, and Perplexity for your priority queries. Being present in AI answers builds recognition that pays off in later clicks, so visibility in AI responses is now a core metric. If you cannot see your presence in AI answers, you cannot manage or improve it.