AI Search Visibility: Getting Cited When ChatGPT and Perplexity Answer for You

A growing share of your buyers' research now happens inside answer engines — ChatGPT with browsing, Perplexity, Google's AI Overviews — where the "result" is a synthesised paragraph with a handful of citations. Brands are discovering their beautifully-ranked pages summarised without attribution while a competitor's dry spec table gets the citation link. The optimisation discipline forming around this (call it GEO — generative engine optimisation) is real, measurable, and overlaps classic SEO less than the agencies retooling their decks suggest.

What answer engines reward

Citable claims, not vibes. Engines quote sentences that assert something checkable: numbers, dates, named comparisons, stated trade-offs. "Fulfilment costs typically run 12–18% of revenue for mid-size EU brands" is quotable infrastructure; "we optimise your fulfilment journey" is filler the model skips. Audit your money pages for sentences a careful writer could cite — most brand copy has none.

Structured data and entity clarity. Schema markup (Organization, Product, Article, FAQ), consistent NAP-style entity facts, and unambiguous "what is this company" statements give engines the graph edges they need to attribute confidently. Engines hedge about entities they cannot resolve — and hedged entities do not get recommended.

Answer-shaped pages. The formats that dominate citations look like this blog deliberately does: a question in the title, a direct answer in the first hundred words, structured sections with scannable claims, honest trade-offs (engines prefer content that states limits — it patterns as trustworthy), dated freshness. Comparison tables get lifted wholesale; write the table you want quoted.

Presence where models already look. Citations skew toward sources models treat as reference-grade: documentation, detailed guides, community answers, industry publications. A thin brand blog re-summarising the category loses to your own genuinely expert piece — the EPR guide, the VAT explainer — which is the content that earns both the citation and the client.

What legacy habits miss

Keyword-density-era writing actively hurts (models summarise past it); link-building matters less than claim quality; and rank-tracking alone measures the wrong funnel — add answer-engine monitoring: run your category's buying questions through the major assistants monthly, log who gets cited, and treat "cited for the question we monetise" as the KPI. Referral traffic from assistants is already visible in analytics if you segment for it; it converts absurdly well, because the assistant pre-qualified the visitor.

The honest strategic read

Answer engines compress the consideration set from ten blue links to two or three named options. Being one of them is winner-take-most; being summarised anonymously is the new page two. The work — real expertise, published in citable form, marked up properly — is slower than buying ads and compounds in a way ads never did. This site's own insights section is the strategy, practised in public; the fifty articles around this one are the evidence.