AI engine readiness — being cited by Perplexity, ChatGPT, Claude, Google AIO
What we know in 2026 about what AI engines actually cite. The five-week plan to go from launch to first citation.
AEO (Answer Engine Optimisation) is the practice of getting your pages cited when AI engines answer questions. We've watched dozens of CW-2 alumni go from launch to first Perplexity citation. The patterns are consistent. This guide is everything we've learned about what works, what doesn't, and a five-week plan to get from a freshly-deployed site to citations.
What gets cited (in priority order)
1.Pages with clean structured data (JSON-LD)
The single highest-leverage move. AI engines preferentially parse pages where 'this is a Service from this Organization at this price' is machine-readable without ambiguity.
Add Person/Organization to your homepage, Service to /services pages, Article to blog posts, FAQPage to anywhere with Q&A.
2.Direct answer in the first 50 words
AI engines look for an answer near the top. A 300-word intro before the answer means your page doesn't get used.
Rewrite your top pages: lead with the answer, expand into the why.
The reference-page test
Read your page aloud. If you couldn't pull a one-sentence answer from the first paragraph, AI engines can't either. Rewrite.
3.Named-entity density
Specific people, products, places, tools mentioned by name. AI engines connect 'Claude Code' to your page about Claude Code workflows.
Don't write 'a popular AI coding tool' — write 'Claude Code'. Specifics beat abstractions for both readers and AI.
4.Recent dateModified
Fresh content gets weighted higher. Update dateModified when you meaningfully revise (real edits, not just touching the timestamp).
This is a strong signal Perplexity uses; weaker but present in Google AI Overviews.
5.Numbers, dates, prices, specifics
'In Q1 2026, three SG fintechs raised Series A funding' beats 'some fintechs raised money'.
Specifics are citable. AI engines prefer pages that say something concrete.
6.Authoritative inbound links
Same SEO principle, narrower interpretation. AI engines weight 'authoritative in this niche' higher than 'authoritative in general'.
Build links from sites your audience trusts. One link from a respected industry blog > 50 from random directories.
What doesn't work
- •Keyword stuffing in meta tags.
- •Hidden text or cloaking — penalty risk.
- •AI-generated content that reads AI-generated. (Irony.)
- •10,000-word essays where the answer is in paragraph 47.
- •Aggressive interstitials, popups, slow load times.
- •Buying GEO services from agencies — most are AEO in a new wrapper, occasionally worse.
Five-week plan from launch to first citation
1.Week 1 — foundations
Get the table stakes right.
- •Meta tags on every page (see /templates/meta-tags-deep).
- •Sitemap + robots + llms.txt deployed.
- •Search Console + Bing Webmaster verified, sitemap submitted.
- •Lighthouse SEO category at 100.
2.Week 2 — structured data
Add JSON-LD everywhere it adds meaning.
- •Person or Organization on homepage.
- •Service on each service page.
- •FAQPage on pages with real Q&A.
- •Article on every blog post.
- •BreadcrumbList on deep pages.
3.Week 3 — rewrite top 5 pages with the direct-answer pattern
For each of your top 5 pages: read it aloud. Pull the answer into the first 50 words. Move context after.
This single pass typically lifts citation likelihood more than any other content change.
4.Week 4 — start publishing topic content
One blog post per week minimum. Topics: things your customers ask. Each post: direct answer, named entities, specific numbers/dates, JSON-LD Article schema.
Internal-link new posts from existing pages. Bidirectional.
5.Week 5+ — monitor citations and iterate
Test queries weekly across Perplexity, ChatGPT search, Claude, Google.
- •Ask your customers' actual questions.
- •Note whose pages get cited.
- •Reverse-engineer the structure of cited pages.
- •Adjust your own content shape based on what wins.
How to monitor citations
Tools exist; most have free trials. Don't pay until you've confirmed you're getting cited.
- •https://www.tryprofound.com/ — tracks brand mentions across major AI engines.
- •https://www.athenahq.ai/ — focused on AI search visibility.
- •https://otterly.ai/ — generative engine monitoring.
- •Free: manually query Perplexity / ChatGPT (with browsing) / Claude weekly with your customer questions. Note when you appear.
Realistic timeline
Most CW-2 sites start showing in Perplexity citations within 4-8 weeks of consistent publishing. ChatGPT search and Google AIO are slower (8-16 weeks). Don't lose faith at week 3.
Troubleshooting
I've been doing everything right for 12 weeks and still no citations.
Three likely causes. (1) Content thinness: do you actually say something specific others don't? Generic content doesn't get cited regardless of structure. (2) Backlinks: AI engines weight authoritative inbound links. Build 5-10 from respected sites in your niche. (3) Wrong queries tested: try more specific customer-language queries, not industry jargon.
I'm being cited but always in a list with 10 others.
That's normal. Citation in a list of 10 is still distribution. Over time, distinctive positioning (your specific service, unique data) lifts you from list member to primary citation.
I'm being cited but with outdated information.
Update dateModified on the page. Add a clear 'last updated: 2026-XX-XX' visible in the body. Resubmit the URL in Search Console (URL Inspection → Request indexing). Citations refresh over 2-6 weeks.
Want to do this with us in the room?
Bring your real project to a full-day workshop and leave with it shipped.
See the workshops