In the previous article we covered how AI engines decide who to recommend. This one turns that into a playbook. The moves have a natural order. Earlier ones unlock the later ones.
1. Get reachable (the prerequisite)
None of the rest matters if the engine can't fetch your page.
- Confirm AI crawlers are allowed: both in
robots.txtand at your CDN. (A CDN can silently 403 them even when robots.txt says yes. See is your CDN blocking AI crawlers?.) - Decide which bots you want across the train / index / answer lifecycle. At minimum allow the index and answer bots.
- Make key content server-rendered, not locked behind JavaScript a crawler won't execute.
2. Get parseable (structure for retrieval)
Retrieval works on passages, so write so an engine can lift a clean answer:
- Answer-first. Lead a section with the direct answer, then explain. Don't bury it.
- Question-style headings that match how people ask ("How much does X cost?").
- Schema markup: Product, FAQPage, Organization, Article, so machines know what each thing is.
- Self-contained chunks. One section should answer one question without needing the rest of the page.
3. Get specific (give them something quotable)
Engines favour verifiable, numeric, named claims. Replace vague marketing lines with concrete ones:
- ✗ "trusted by leading brands" → ✓ "used by 40 D2C beauty brands across 6 markets"
- ✗ "fast setup" → ✓ "live in under 5 minutes, no developer required"
Specific claims get pulled into answers; vague ones get skipped.
4. Get corroborated (the lever brands underuse)
This is the big one most teams miss. AI engines lean heavily on third-party authoritative sources, not just your own site. A claim echoed on Reddit, G2, review sites, and press reads as corroborated; the same claim only on your own page reads as a lone assertion.
- Pursue earned mentions and reviews.
- Get into independent "best [category]" roundups: for buyer queries these get cited far more than your own pages (here's why).
- Keep your brand facts consistent everywhere they appear, so the web tells one coherent story about you.
5. Build entity strength (the slow compounding one)
The engines reward brands that are well-established entities: a Wikipedia/Wikidata presence, steady brand search volume, consistent co-occurrence with your category. This is slow, but it's what makes a model "know" you well enough to recommend you unprompted.
6. Stay fresh
There's roughly a three-month citation cliff: content decays if you publish and forget. Refresh your most important pages on a cadence; updated, accurate pages win live retrieval over stale ones.
Order matters: reachable → parseable → specific → corroborated → entity → fresh. Skipping to "write more content" before you're reachable and parseable is the most common wasted effort in GEO.
Work the list top to bottom and you move from invisible to citable. The part teams can't do by eye. Knowing which of these is actually holding you back, on which engine. Is what measuring your AI visibility (and Buffy Intel) is for.