Answer engines do not return ten blue links. They synthesise a response and cite a handful of sources. Being one of those sources is a different discipline from ranking — here is the method we actually use.
A meaningful share of informational queries now end without a click. The user asks, the engine answers, and two or three sources are cited underneath. If you are not one of them, you were not in the conversation at all — regardless of where you rank on the traditional results page.
What answer engines actually reward
Having audited a lot of citation data across four engines, the pattern is consistent. Cited sources tend to share four properties.
- They answer the question directly, near the top, in a self-contained passage.
- They contain verifiable specifics — numbers, dates, named sources — rather than general assertion.
- They carry clear entity signals about who is speaking and why that party is credible.
- They are structurally unambiguous: proper headings, marked-up questions, clean HTML.
Answer-first structure
Put a 40–60 word direct answer immediately beneath each question heading, before any context or preamble. This is not a summary of the section — it is the answer, written to be lifted out and still be accurate on its own. Everything else on the page supports it.
This is uncomfortable for writers trained to build toward a conclusion. Do it anyway. An engine extracting a passage does not read the run-up.
Entity clarity
Answer engines need to establish who is making a claim before they will repeat it. That means consistent organisation naming across the web, Organization and Person markup with sameAs links to authoritative profiles, named authors with real credentials, and an About page that states plainly what you do and since when.
Give them something only you have
The single highest-leverage AEO tactic is publishing first-party data. Aggregated, anonymised numbers from your own client work cannot be synthesised from anywhere else, so an engine answering a question in that territory has to cite you or omit the specific.
This is also why generic content is losing so quickly. If a passage can be reconstructed from twenty other pages, the model does not need yours.
Measure citations, not positions
There is no position to track. Sample your priority questions across ChatGPT, Perplexity, Gemini and Google AI Overviews on a fixed schedule, record whether you were cited and in what context, and watch the trend. Then triangulate with the proxies that do move: branded search volume, direct traffic and AI Overview impressions in Search Console.