To appear in AI search results, you need to win on four levers that every engine shares: retrievability (the engine can fetch and parse you), extractable structure (your answers lift cleanly), corroboration (independent sources confirm your facts), and topical authority (you cover your subject deeply enough to be trusted). ChatGPT, Perplexity, Gemini, AI Overviews and Copilot differ in how they retrieve and how visibly they cite, but sources that win across all of them win the same four ways. This is the umbrella playbook: the levers, the per-engine nuances, a 30-day starting sequence, and the one metric that tells you it is working. It sits on the foundations of answer engine optimization and generative engine optimization, and links into the per-engine guides where the details diverge.
The four levers that work everywhere
Retrievability. Before any engine can cite you, it has to fetch and parse you. That means crawler access for AI user agents in robots.txt, content that renders without depending on client-side JavaScript, clean heading hierarchies, structured data where it fits, and an llms.txt file pointing machine readers at your canonical pages. Our AI search optimization checklist is the full pass. Extractable structure. Engines compose answers from content they can lift: definitions stated in the first sentence, comparisons in honest tables, one question answered per section. The LLM SEO formats guide catalogs what gets quoted. Corroboration. A claim that appears only on your site reads as marketing; the same claim in independent coverage, communities and directories reads as consensus, and engines cross-reference before they commit facts to an answer - the selection mechanics are in our AI citations deep dive. Topical authority. Engines favor sources that cover a subject deeply and consistently over isolated landing pages, so clusters beat one-offs. Get these four right and every engine below becomes winnable; skip any one and the others carry less.
Engine-by-engine nuances
| Engine | Retrieval (observable) | What to emphasize |
|---|---|---|
| ChatGPT | Answers from training plus web browsing on queries that need it; cites mainly when it browses | Corroboration and entity consistency, so both the model's memory and its browsing agree on your facts |
| Perplexity | Live retrieval per query; numbered citations on every answer | Freshness and direct answers; the most measurable engine - see Perplexity SEO |
| Gemini | Can ground answers in Google Search results | Classic Google rankings plus liftable structure - see Gemini SEO |
| AI Overviews | Composed inside Google Search from indexed pages; citation display varies by query | Indexing health and extractability - see our AI Overviews guide |
| Copilot | Grounds answers in Bing's index with linked sources | Bing indexing health, often the forgotten crawl surface |
Two things the table implies are worth saying outright. First, classic SEO is still the foundation: Gemini, AI Overviews and Copilot all retrieve from search indexes, so a page the crawlers cannot reach or will not rank is invisible to most of the AI surface too. Second, the engines are converging rather than diverging - each one rewards the same retrievable, extractable, corroborated, authoritative source, so the right strategy is one program measured per engine rather than five separate programs.
The 30-day starting sequence
Days 1 to 7: baseline. Write down the 30 to 50 questions your buyers actually ask, put them to each engine, and record every answer and every cited source - the AI visibility audit is the full method. You cannot improve an answer you have not read, and the baseline is what makes later movement provable. Days 8 to 14: the technical pass. Fix retrievability end to end - crawler access, rendering, structured data, llms.txt, entity-consistent facts - because these problems block everything downstream and are usually the cheapest to fix. Days 15 to 30: winnable gaps. From the baseline, pick the questions where current citations look displaceable - thin pages, stale content, generic sources - and ship liftable content against them: definition pages that answer in the first sentence, comparisons with real tables, spec answers. Then put the measurement loop on a weekly cadence and let the program compound. Expect first movement in weeks on live-retrieval engines and slower movement where index refresh and authority accrual set the pace.
After day 30, the program settles into three standing loops. The measurement loop keeps the weekly question runs going and flags every answer where a competitor displaced you. The content loop ships against the gaps measurement surfaces, in priority order of commercial value times displaceability. The corroboration loop places your core facts in independent sources on an ongoing basis - coverage, communities, directories, partner pages - because that lever moves slowest and compounds longest. Teams that stall usually stalled on the loop cadence rather than the tactics; the ones that compound treat this like a weekly operating rhythm with an owner.
Measurement: citation share
Everything in this playbook is cadence work - weekly question runs, content shipped against measured gaps, technical checks kept green - which is why it suits agents better than heroics. On AstroFabric, the AI Visibility agent runs the question set across engines, tracks citation share against your baseline, and feeds winnable gaps into your content queue, inside a platform built for growth, revenue and digital operations; the AI visibility solution maps the standing program.
Frequently asked questions
How do you appear in AI search results?
Win four levers every engine shares: retrievability (crawler access, clean rendering, structured data), extractable structure (direct answers, honest tables), corroboration (independent sources confirming your facts), and topical authority (real depth on your subject). Then measure citation share per engine.
How do you get cited by AI?
Engines cite sources they can retrieve, lift cleanly, and verify. State the answer in your first sentence, structure pages so sections answer one question each, get your facts corroborated in independent places, and keep classic SEO healthy since most engines retrieve from search indexes.
Do you need a different strategy for each AI engine?
One program, measured per engine. The four levers carry everywhere; the engines differ at the margins - Perplexity rewards freshness, Gemini and AI Overviews lean on Google rankings, Copilot on Bing indexing, ChatGPT on corroboration. Tune emphasis per engine rather than running five programs.
Does classic SEO still matter for AI search?
Yes - it is the foundation. Gemini, AI Overviews and Copilot retrieve from search indexes, and Perplexity and browsing ChatGPT fetch from the open web, so crawlability, indexing health and authority decide what the engines can select from in the first place.
How long does it take to show up in AI answers?
On live-retrieval engines like Perplexity, a crawlable page can be cited within days. On index-backed surfaces like AI Overviews and Gemini, movement follows indexing and ranking timelines - weeks to months. The compounding is the payoff: source trust, once earned, persists across answers.
Sources
- Google Search Central - AI features and your website
- llms.txt - the proposed standard for LLM crawler guidance
- GEO: Generative Engine Optimization (the original research paper)
Every playbook on this blog ships as a runnable mission.
Open a workspace and the playbook library is waiting - describe the outcome and the agents carry it end to end, on your plan's monthly credits.