AI engines (ChatGPT, Claude, Perplexity, Gemini) now answer many of the queries that historically went to Google. When your brand or product is named in an AI response, you get the modern equivalent of a top-three search result — a citation that drives both traffic and trust. The AI Visibility Tracker measures how often your brand is cited across a defined prompt set, week over week, so you know whether your AEO investments are working. This guide covers the methodology, the prompt design, the measurement cycle and how to interpret the results.
AI Visibility is the rate at which your brand, product or content is cited in answers from ChatGPT, Claude, Perplexity, Gemini and other AI engines, measured across a stable, representative prompt set. Unlike traditional SEO rankings (which Google reports through Search Console), no AI engine publishes citation share data — you have to measure it yourself by querying the engines and counting your appearances.
The prompt set is the foundation of measurement. A bad prompt set produces meaningless data; a well-designed one tracks the queries your customers actually ask. Build the prompt set across three layers:
Direct queries about your brand: "Tell me about [Your Company]", "Is [Your Company] legitimate?", "What does [Your Company] do?". These set a baseline for your branded AI presence.
Generic category queries where you want to be cited: "Best [your category] tools for [audience]", "How do I solve [problem your product solves]?", "Alternatives to [dominant competitor]". These reveal your share-of-voice in the AI category conversation.
Real customer-language problem statements without naming your category: "My team keeps losing track of [problem]", "I need to [job to be done]". These reveal whether your content is being surfaced for jobs-to-be-done, not just brand-aware queries.
Aim for 30-100 prompts total. Smaller is acceptable for narrow niches; larger isn't always better because every prompt needs to be re-run weekly, and engine API costs add up.
Before optimising, measure where you stand. Run the full prompt set against all four engines and record:
Typical baselines for new SaaS brands: 5-20% citation rate on category prompts, 60-90% on brand prompts. Established category leaders run 40-80% on category prompts.
Run the full prompt set weekly, same day, same time. Engine results vary day to day; weekly cadence smooths the noise. Record the data in a spreadsheet with one row per (prompt, engine, week) so you can chart trends.
What the numbers mean:
| Pattern | Diagnosis |
|---|---|
| Citation rate rising on category prompts | AEO investments working — content is being indexed and weighted |
| Citation rate flat for 8+ weeks | Your category authority isn't growing — invest in original research, expert content, or PR |
| Citation rate falling | Either competitors caught up, or you've drifted out of fresh-indexing windows — refresh content cadence |
| Cited but description wrong | Your structured data or canonical positioning is sending mixed signals — see how-to-fix-wrong-attribution |
| One engine high, others low | Your content profile fits that engine's source mix — diversify sources for broader coverage |
The top correlates of AI citation rate:
Use weekly data to prioritise content investment. If you're cited at 40% on prompt A but 5% on prompt B, prompt B is the under-served opportunity. Build content specifically to answer prompt B better than current sources, get external mentions on the same topic, and measure the impact over the next 6-12 weeks.
Run prompts across ChatGPT, Claude, Perplexity and Gemini. Track citation rate, position and competitive share.
Open AI Visibility Tracker →