Access
Can the system reach your page?
- You never appear in any AI answer: Check robots.txt and noindex tags on key pages. Verify AI crawlers (GPTBot, ClaudeBot, Bingbot) are not blocked.
- You rank in classic search but never get cited: Check client-side rendering. If content loads via JS, AI crawlers may not see it.
- You appear in Google AI Overviews but not ChatGPT or Perplexity: Check Bing indexation separately. ChatGPT and Perplexity use Bing's index, not Google's.
Full checklist → Retrieval
Does your page get pulled into context?
- You rank but rarely get cited: Check whether you have pages that directly answer sub-questions, not just the main keyword.
- A competitor is consistently cited instead of you: Check whether you have a definition page, comparison page, or criteria page for that topic.
- Citations are inconsistent week to week: Check internal link structure. Hub-to-spoke linking signals topical hierarchy to retrieval systems.
Full playbook → Extractability
Can the model lift a correct chunk?
- You get cited but the AI answer misrepresents you: Check whether each section opens with a direct answer in the first 1-2 sentences under the header.
- Your ideas appear but are paraphrased poorly: Check for orphan pronouns. Paragraphs with "it/this/that" without a noun break when excerpted.
- Competitors get cleaner quotes than you: Check whether you have explicit quote blocks: 2-4 sentences designed to be lifted as a standalone passage.
Full spec → Name
Does your brand attach to the claim?
- Your ideas appear but your brand name doesn't: Bind brand to definition once per major claim: "At [Brand], we define X as..." Use sparingly.
- Your framework is described generically: Check whether the framework name is stable and consistent across pages. Renaming breaks entity association.
- No authorship signal in the AI answer: Check whether author name appears in byline, bio, and near at least one major definition on the page.
Full guide → Authority
Do you keep showing up as sources rotate?
- You used to show up but stopped: Check whether you track the same prompt set weekly. Drift is invisible without a baseline.
- You dropped after a competitor published new content: Check whether your best quote blocks still exist unchanged. Casual edits can break trusted patterns.
- Strong on-site content but thin off-site presence: Check third-party corroboration. Directories, podcasts, and guest posts stabilize your position.
Full checklist → Measurement
Are you tracking the right signals?
- You're using rankings as your AI visibility proxy: Switch: track cited URLs and accuracy score (0 = wrong, 1 = partial, 2 = correct) per prompt per surface.
- You don't know which ARENA step is broken: Run 20+ prompts grouped by intent cluster. The pattern of absence tells you which step is failing.
- You can't tell why a competitor replaced you: Compare the chunk they're being quoted from to your equivalent passage. It's usually a writing quality gap.
Full template →