Checklists are only useful if they are specific enough to act on. This one is built from the audit process I run at Salterra for clients who want to understand and improve their AI citation presence. Work through it top to bottom, or use it as a gap analysis against your current site.
Entity Foundation
Before content structure or technical SEO, AI engines need a clear, confident picture of who you are. These items build that foundation:
- Your business name, category, and description are identical across Google Business Profile, LinkedIn, Crunchbase, Wikipedia/Wikidata, and any major industry directory.
- Your About page reads as a factual entity statement: who you are, what you do, where you operate, how long you have been doing it. No marketing adjectives, just facts.
- Founder and key team members have complete LinkedIn profiles with clear titles and specialties.
- If your company qualifies for a Wikipedia entry or Wikidata record, those exist and are accurate.
- Your Google Business Profile is fully completed: services list, description, hours, Q&A, regular posts.
This section is where I start every AEO engagement, before touching content or schema, because entity confusion undermines everything built on top of it. If Google Business Profile lists you as one category and your About page describes something adjacent, an AI system pulling from both sources has no clean way to resolve the discrepancy in your favor — it just treats you as a less reliable source. I have seen companies with genuinely strong content sit invisible in AI answers because their entity signals contradicted each other across five different profiles.
Content Structure
AI engines extract answers from well-structured content. Check each high-priority page:
- The page answers its target question directly in the first 100 words — not after three paragraphs of scene-setting.
- H2 headings are written as questions or clear answer phrases, not as vague categories (‘Overview,’ ‘More Information’).
- FAQ sections exist on service and topic pages, with complete, direct answers (150–400 words each for complex questions).
- Lists are used for steps and criteria; paragraphs are used for explanations — not the reverse.
- No page is under 800 words for a topic that merits comprehensive treatment.
- Every claim of fact is either self-evidently true, sourced, or clearly marked as opinion — AI systems are cautious about surfacing unsupported assertions on topics with any stakes.
The most common failure I find here is not bad writing — it is structure that buries the answer. A page can have excellent, accurate information and still fail an AEO audit because the direct answer is paragraph four instead of paragraph one. I rewrite the opening of a page far more often than I rewrite its actual substance.
The full content framework is in AEO content strategy.
Schema and Technical Signals
- Organization schema on the homepage with legalName, url, logo, sameAs links to all major profiles.
- Person schema for founder and key team member pages.
- FAQPage schema on all FAQ sections.
- Article schema on blog and editorial content, with author entity linked.
- BreadcrumbList schema on all interior pages.
- Pages are fully crawlable — no content-blocking JavaScript without proper fallback, no index-blocking robots meta on intended citation targets.
- Site loads in under 3 seconds on mobile (Core Web Vitals green).
- llms.txt file exists at the domain root with a factual site summary and key page list.
Schema does not create trust on its own — it makes trust machine-readable. A site with strong entity signals and clean content but no structured data is making an AI system do more inferential work to understand what it is looking at. That is not necessarily disqualifying, but it is a disadvantage against a competitor who has marked everything up explicitly. I treat schema completeness as a baseline hygiene item, not a differentiator — everyone serious about this should have it, so having it just keeps you in the running rather than putting you ahead.
Off-Site Entity Presence
- Your brand name appears in at least one editorial source (trade publication, local press, industry blog) independent of your own content, per quarter.
- Review profiles (G2, Yelp, Google, or vertical equivalent) have recent reviews with descriptive text — not just star ratings.
- You have been cited as a source in at least one third-party piece in the past six months (as a quote, data source, or reference).
- Podcast appearances, conference talk transcripts, or webinar recordings with indexed transcripts exist for key personnel.
This is the section clients most often want to skip, because it is the slowest to show results and the hardest to fully control. I do not blame them for the instinct, but it is usually where the real ceiling is. A site can max out every on-site item on this checklist and still be invisible in AI answers if nothing outside the domain corroborates who they are. Off-site presence is the evidence an AI system uses to decide your on-site claims are credible.
At Salterra, we track this off-site presence quarterly as part of ongoing AEO retainers — not just backlinks, but named entity mentions in indexable contexts.
Original Data and Unique Content
- Your site has at least one piece of original data — a survey, benchmark report, or proprietary analysis — published and kept current annually.
- That data page follows citation-friendly formatting: headline stat, methodology note, dated attribution, stable URL.
- Your content makes specific, defensible claims rather than generic assertions. Every opinion is accompanied by reasoning; every statistic is attributed.
- Your data page is built to be cited cleanly — a journalist, blogger, or AI system referencing it should be able to pull one clear number with attribution, not have to synthesize it from a dense report.
AI Engine Testing
- You have tested your top 10 target queries in ChatGPT (browse mode), Perplexity, and Google AI Overviews in the last 30 days.
- Results are logged: which queries surface your brand or content, which cite competitors, which cite no one relevant.
- Gap queries (where you should appear but do not) have been mapped to specific content or entity gaps.
Manual testing feels crude next to the rest of this checklist’s more structural items, but it is the only ground truth you have. Everything else on this list is an input; this section is the output check. I run the test queries logged out, in a fresh session, because personalization and history can make your own results look better than what a new searcher would actually see.
The measurement process is covered in detail in measuring AI visibility.
Ongoing Maintenance
- AI engine tests run on a fixed cadence — monthly for competitive categories, quarterly for stable niches.
- Entity profile updates are reviewed semi-annually — business description, services list, team profiles.
- New content is evaluated against the direct-answer standard before publishing: does this page answer its target question in the first paragraph?
Prioritizing When You Can’t Do Everything at Once
Most sites I audit fail on more items than they pass, and trying to fix everything simultaneously is how AEO work stalls out. I tell clients to fix Entity Foundation and Content Structure gaps first — they are the cheapest to fix and the most foundational, since AI systems that cannot confidently identify who you are will not cite you regardless of how good the rest of the checklist looks. Off-site presence comes next, worked on continuously rather than in a single sprint, because it cannot be rushed without looking manufactured. Original data projects I schedule last, since they take the longest to plan and produce, but they tend to be the highest-leverage item once everything else is in place — a genuinely original statistic gets cited in ways that well-written explainer content rarely does.
Using This Checklist
Run through each section and mark items as done, in progress, or not started. The not-started items in the Entity Foundation and Content Structure sections are usually where the biggest gains are. Off-site presence is often the longest-timeline item — build it consistently rather than in bursts, and expect it to take quarters, not weeks, to fully mature.
The full strategic framework that this checklist maps to is in the AEO overview — it explains the reasoning behind each layer and how they interact.