Answer engine optimization (AEO) went from niche buzzword to must-do in 18 months. Every CMO has it on a slide. Almost none of them have a concrete execution plan.
This is that plan. 40practical items across five categories — technical foundation, schema, content, entity signals, and monitoring — that cover what actually moves the needle for citations in ChatGPT, Claude, Gemini, Perplexity, and Deepseek. Run it top to bottom, check boxes as you go, and you'll have a site the AI answer engines can actually cite.
We'll refresh this list every year as platforms evolve. For the 2026 edition, we've added items reflecting Deepseek's emergence, Google's expanded AI Overview surface, and the llms.txt standard reaching broad adoption.
In this checklist
1. Technical foundation (10 items)
If AI crawlers can't read your site, nothing else matters. Start here.
- 1
Server-rendered HTML
Hero copy, service descriptions, and key content appear in View Page Source — not hydrated by JavaScript. If you're on React/Vue/Angular, use SSR or SSG (Next.js, Nuxt, Astro).
- 2
robots.txt allows AI crawlers
Explicitly allow GPTBot, ClaudeBot, PerplexityBot, Google-Extended, and Applebot-Extended. Blocking them by default = invisible to AI.
- 3
llms.txt at /llms.txt
Markdown summary of your business, key pages, and preferred framing. Anthropic, OpenAI, and Perplexity have signaled support; Google is moving in the same direction.
- 4
sitemap.xml current and submitted
Every indexable page listed with current lastmod dates. Submit to Google Search Console and Bing Webmaster.
- 5
Core Web Vitals green on mobile
LCP under 2.5s, INP under 200ms, CLS under 0.1. AI crawlers weigh performance as a signal of site quality.
- 6
Mobile-first rendering
Google's index is mobile-first. AI crawlers follow. If your mobile experience is broken, AI sees a broken site.
- 7
HTTPS with valid certificate
Non-negotiable. AI crawlers skip HTTP-only domains and sites with cert errors.
- 8
Canonical tags on every page
Prevents duplicate-content confusion. Especially important if you have pagination, filters, or marketing variants.
- 9
Zero broken internal links
404s on internal links signal site neglect. Run a crawl quarterly with Screaming Frog or Sitebulb.
- 10
Clean redirect chains
No 301 → 301 → 301 chains. Each hop loses signal. Collapse to single redirects.
2. Schema & structured data (8 items)
JSON-LD is how you tell crawlers what your content actually is — Organization, Service, FAQ, Article, Person. Without it, AI guesses.
- 1
Organization schema in root layout
Include name, URL, logo, sameAs (social profiles), and founder/Person reference. This is the foundation entity.
- 2
Person schema for key people
Founder, principal consultants, authors. Include jobTitle, worksFor, sameAs (LinkedIn, GitHub, etc.).
- 3
Service schema on service pages
Each offering gets its own Service schema with provider, areaServed, and priceSpecification where public.
- 4
FAQPage schema on pages with Q&As
Every FAQ block gets wrapped in FAQPage JSON-LD. Unlocks rich results and strong AEO citation signals.
- 5
Article schema on blog posts
Headline, datePublished, author (Person), publisher (Organization), mainEntityOfPage. Required for news/article crawlers.
- 6
BreadcrumbList schema
Helps crawlers understand site hierarchy. Adds breadcrumb trails in SERPs.
- 7
Schema passes Google Rich Results Test
Run every key page through search.google.com/test/rich-results. Zero errors, zero warnings.
- 8
@graph wrapper for multi-entity pages
When a page references multiple entities (e.g., Organization + Service + FAQPage), wrap them in @graph to show relationships.
3. Content & entity clarity (9 items)
AI models reward specificity, depth, and original material. They punish generic keyword-stuffed pages. Rewrite accordingly.
- 1
Specific numbers on service pages
"We improved conversion 34% on a $40M ARR SaaS" beats "we deliver results." Every service page needs at least one concrete client outcome with a number.
- 2
Named people on service and case study pages
"Chris Igbojekwe led the migration" beats "our team executed the work." Ties work to Person schema and entity graph.
- 3
Opinions most competitors won't state
AI models cite confident positions. Bland pages get summarized without attribution. Pick fights; take stances.
- 4
At least 3 pieces of original data
Benchmarks from your client work, survey data, proprietary frameworks. Unique material is citation-worthy by default.
- 5
FAQ section on every commercial page
6-10 buyer-language questions per page. FAQPage schema attached. Both ranks in Google PAA and gets cited by AI.
- 6
Entity-clear language (not keyword-stuffed)
"Seed App is a boutique consultancy in San Diego" beats "top-rated SEO-friendly award-winning agency." Clear > clever.
- 7
2+ long-form pillar posts (2,000+ words)
Cornerstone content on your core topics. Internally link service pages to them. Builds topical authority AI models can detect.
- 8
Case studies with specific client names
Where permitted, name real clients. Anonymous case studies are treated as weaker signals by both Google and AI.
- 9
No AI-generated filler content
AI models detect AI-generated fluff and deprioritize it. Write it yourself, or have a human edit aggressively.
4. Entity signals (7 items)
AI models build entity graphs from external sources. If you're barely represented, the model has shallow context to cite you confidently.
- 1
LinkedIn company page complete and active
Full description, logo, industry, size, specialties. Post at least monthly. Dormant pages signal dormant companies.
- 2
Key people have coherent LinkedIn profiles
Founder and leadership with consistent bios, company attribution, and activity. AI cross-references these for Person entity verification.
- 3
Crunchbase profile accurate and current
Company description, founding year, headquarters, team. Crunchbase is a major entity source for AI models.
- 4
Listings on 3-5 credible directories
Clutch, G2, Capterra, DesignRush for agencies; industry-specific for other verticals. Avoid low-quality link-bait directories.
- 5
Wikipedia page (if eligible)
Only if you meet notability standards. Don't force it. When it fits, it's the single strongest entity signal available.
- 6
2-3 credible press mentions per year
Trade publications, industry podcasts, notable blogs. Quality matters more than quantity.
- 7
Consistent name, address, phone everywhere
Footer, contact page, LinkedIn, Crunchbase, Google Business Profile. Discrepancies reduce entity confidence.
5. Monitoring & iteration (6 items)
AEO isn't a set-and-forget project. Without measurement, you have no idea what's working.
- 1
Defined monthly query set (25-50 queries)
Mix of brand queries, category queries, problem queries, and competitor queries. Test the same set every month.
- 2
Test across ChatGPT, Claude, Gemini, Perplexity, Deepseek
Each platform cites differently. What works in ChatGPT may not work in Gemini. Platform-specific insights matter.
- 3
Citation log with screenshots + URLs cited
Track which of your pages get cited vs. which get summarized without attribution. The cited pages are winning; the others need work.
- 4
Google Search Console monitored weekly
Rankings, impressions, CTR on priority terms. AI visibility starts with traditional search visibility — they share a foundation.
- 5
Refresh pillar content every 6 months
AI models and Google both weigh freshness. Updated dates, updated stats, expanded sections. Don't let pillars rot.
- 6
Quarterly review of new AI platforms
2025 gave us Perplexity and Claude as major players; 2026 added Deepseek. Check each quarter for new platforms worth monitoring.
How to prioritize if you can't do everything at once
40items is a lot. Here's the triage order I'd run with a client who can only ship in phases:
- Phase 1 (week 1-2):Technical foundation + schema. These are non-negotiable. If AI crawlers can't read or understand your site, everything else is wasted effort.
- Phase 2 (week 3-6):Content restructuring + FAQ blocks on commercial pages. Rewrite for entity clarity, add specific numbers, publish one long-form pillar.
- Phase 3 (week 7-10):Entity signal cleanup. LinkedIn, Crunchbase, directories, press. This compounds for months after.
- Phase 4 (ongoing):Monitoring setup. Monthly query sets, citation log, quarterly reviews. This is how you stay ahead.
What to expect after shipping the checklist
First citations typically appear in Perplexity within 2-3 weeks of shipping the foundational fixes. ChatGPT and Claude follow at 30-60 days. Gemini at 60-90 days. Deepseek is still a wildcard — newer, signal patterns still forming.
If you're at 90 days post-implementation with no citation movement, something's wrong. Usually it's weak entity signals (the external graph) or thin content (the internal signal). Go back to Phases 2 and 3 and push harder.
Want us to ship this checklist for you?
We implement the full AEO checklist as part of our Website Redesign + AI Visibility engagements. Strategy, design, SEO, AEO, and citation monitoring in one team. Engagements start at $25,000.