Generative engine optimization has enough moving parts that things get skipped. This GEO checklist covers the twenty checks we run on every Aethon engagement, grouped into the four layers that decide whether AI recommends you.
One: a single sentence stating who you are for, usable by a stranger to route customers. Two: that sentence appears verbatim on your homepage, about page, and top profiles. Three: your three to five winnable buyer situations are documented. Four: each situation has a page that answers it directly. Assistants cannot recommend what they cannot categorize; this layer fixes that.
Five: every money page answers its question in the first two sentences. Six: headings are phrased as buyers phrase questions. Seven: FAQ schema on question pages, Article and Organization schema site-wide. Eight: llms.txt published at your root. Nine: pricing and comparison questions answered honestly somewhere on your site. Ten: no critical content rendered only by JavaScript.
Eleven: review profiles (G2, Capterra, Google, category-relevant) claimed, current, and growing. Twelve: you know the top ten sources Perplexity and Gemini cite for your category. Thirteen: you appear in at least half of them. Fourteen: your descriptions match across every directory and profile. Fifteen: stale or wrong third-party pages about you are identified with correction plans.
Sixteen: a fixed set of twenty real buyer questions exists. Seventeen: baseline mention rates recorded across ChatGPT, Gemini, Claude, and Perplexity. Eighteen: competitor share of voice tracked on the same questions. Nineteen: sampling repeats at least weekly. Twenty: every content or citation fix is tied to the answers it is meant to move, and verified after shipping.
Sixteen or more checks passing: you are ahead of nearly every competitor. Ten to fifteen: real gaps, prioritize layers 3 and 4. Below ten: your AI answers are being written without you. Aethon runs this checklist as a structured audit and ships the fixes layer by layer, measured against your baseline.
A structured set of checks across positioning, content, citations, and measurement that determines whether AI assistants can find, trust, and recommend your brand.
Citation mapping: most teams have never looked at which sources Perplexity and Gemini actually cite for their category, so they optimize pages nobody quotes.
Quarterly for the full checklist, with measurement running continuously, since model updates shift answers between audits.
Layers 1-3 yes, manually. Layer 4 requires repeated sampling across four assistants, which is impractical by hand at meaningful sample sizes.
Yes. It is the backbone of our audit, followed by the remediation work and continuous monitoring that keep the checks passing.