There are three things you have to get right to be visible to AI engines: structured data that tells the model who you are, brand authority across the third-party surface AI engines pull from, and canonical entity naming so your mentions do not fragment. The rest is execution. This guide walks the three in implementation order, using the schema AirPulse itself ships on airpulse.ai as the worked example. For the scoring model behind this, see the six-dimension playbook and how AEO differs from traditional SEO.
1. Structured data, with a real example
Schema.org JSON-LD is how you hand an AI engine machine-readable facts about your brand. Do not ship every schema type Schema.org defines — ship the ones that map to how you want the model to talk about you. The eight types AirPulse currently ships in production are Organization, SoftwareApplication, Service, Article, Person, HowTo, FAQPage, and BreadcrumbList. (HowTo and FAQPage no longer produce Google rich results — Google dropped HowTo in 2023 and FAQ rich results in May 2026; ship them now as semantic signals AI engines parse, not for SERP rich results.)
Here is the Organization block from airpulse.ai, with the @id pattern that lets every other block on the site reference it instead of redeclaring identity:
{
“@context”: “https://schema.org”,
“@type”: “Organization”,
“@id”: “https://airpulse.ai/#organization”,
“name”: “AirPulse”,
“alternateName”: [
“AirPulse AI”,
“airpulse.ai”
],
“foundingDate”: “2025”
}
Three things to copy from this pattern. First, the stable @id — every later block points back to this URL. Second, alternateName covering the casing variants buyers and analysts actually use. This feeds the mention-detection fuzzy pass directly — see section 3. Third, foundingDate, because AI engines anchor a brand on the timeline when they decide whether the information they have is current.
The SoftwareApplication block on the same site declares featureList including ‘AI Visibility Score across ChatGPT, Claude, Perplexity, Gemini, and Copilot’. That phrase is a deliberate engine-naming choice. It lets a model retrieving the page know exactly which engines we monitor without inferring it from prose.
2. Authority signals AI engines actually read
AI engines are risk-averse synthesisers. They prefer to cite sources corroborated across independent pages. Brand authority in 2026 is therefore less about your domain rating and more about distribution of mentions across surfaces a model is likely to ingest:
- Third-party reviews and roundups in industry publications and analyst notes.
- Community discussion that mentions you in context — Reddit threads, Quora answers, niche forums. Reddit is the single most-cited domain across ChatGPT, Perplexity, and Google AI Overviews.
- Comparison pages — vendor-vs-vendor, alternatives-to-X pages.
- Founder and employee bylines on substantive industry posts.
Brand Authority sits at 20% in the audit, second only to Citability itself. That weighting reflects what we see: distribution of mentions across independent surfaces is the most reliable predictor of whether the next AI answer for your category will name you.
3. Entity hygiene: stop fragmenting your own mentions
When AirPulse counts mentions, the algorithm runs an exact pass first and a fuzzy pass second. The fuzzy pass uses a Levenshtein similarity threshold of 0.8 and a ±100-character context window, scoring each candidate with a confidence between 0.0 and 1.0.
In practice, variants close enough to your canonical name — a missing space, a casing flip, an extra ‘AI’ suffix — usually merge. Variants farther away — rebrands, sub-brands, abbreviations — do not. The fix is upstream. Pick a canonical name, list every legitimate variant in your Organization alternateName, and use the canonical form consistently in titles, bylines, and owned media. The cost of getting this wrong is silent: your mentions are real, but your mention count is half what it should be.
4. The audit, in implementation order
Six dimensions, with the weights we apply:
- AI Citability & Visibility (25%) — tracked prompt set across the five engines.
- Brand Authority Signals (20%) — third-party domain mentions, entity consistency.
- Content Quality & E-E-A-T (20%) — author byline, source quality, claim density.
- Technical Foundations (15%) — crawlability for AI crawlers and search bots, render path, status codes.
- Structured Data (10%) — schema coverage and consistency across pages.
- Platform Optimization (10%) — engine-specific patterns. Perplexity, for example, rewards crisp answer-first openers.
Lowest score below 60 sets the priority. Within a dimension, implementation order goes structural, then content, then authority. Structural fixes are deterministic and ship in a day; authority fixes compound over months.
5. Measuring whether any of this worked
Re-run the same prompt set 30 days after the structural changes ship. Three numbers matter: mention rate (did you appear), position rate (did you appear in the first three named recommendations), and sentiment distribution (positive / neutral / negative split). All three feed the Citability score. None of them is the same as click-through.
FAQ
How much schema is enough?
Start with Organization, Article on every post, and BreadcrumbList. Add FAQPage on any page with a real FAQ. Add SoftwareApplication or Service if you sell one. Adding everything Schema.org defines is noise. One caveat: Google removed FAQ and HowTo rich results (HowTo in 2023, FAQ in May 2026), so FAQPage no longer earns a SERP rich result — ship it now only as a semantic signal AI engines parse.
Does Wikipedia presence still help?
Yes, disproportionately. Wikipedia is a high-trust source AI engines prefer. The page has to be earned — do not write it yourself — but it is a real authority lever. Wikipedia consistently ranks among the most-cited domains in LLM answers.
What if my brand name fuzzy-matches a much bigger entity?
Then the canonical form has to be tighter. AirPulse uses ‘AirPulse’ as the canonical and ‘AirPulse AI’ as the alternateName to avoid drift with unrelated ‘pulse’ brands.
— — —
Related AirPulse guides
How to Improve Your Ranking on AI Engines: The AirPulse Playbook
AEO vs Traditional SEO: What Actually Changes Under the Hood
How this was made
How this was made: the draft of this article was generated from AirPulse, our own AI engine optimisation platform, then reviewed and edited by the named author. Product claims about AirPulse are sourced from internal documentation; external claims link to their primary source.
Sources
- AirPulse production JSON-LD (Organization, SoftwareApplication, Service blocks visible in page source) — https://airpulse.ai
- AirPulse audit dimension weights — internal product documentation
- AirPulse mention-detection algorithm — internal product documentation
- Schema.org — https://schema.org/Organization
- Google Search Central — Structured data general guidelines — https://developers.google.com/search/docs/appearance/structured-data/intro-structured-data
- Companion: ‘Introducing next-aeo’ — https://airpulse.ai/insights/introducing-next-aeo-optimize-your-next-js-app-for-ai-visibility
Companion: ‘Where AI Gets Its Sources’ — https://airpulse.ai/insights/where-ai-gets-its-sources-a-practical-guide-to-chatgpt-perplexity-google-ai-overviews