Methodology · updated 2026-10-03

How zitaso measures AI visibility

Transparent method, repeatable runs and every raw answer kept – so each number can be checked.

1. Buying questions, not keywords

For each keyword, zitaso phrases realistic buying questions in the customer’s language, for example “What is the best place to buy Biofinity contact lenses online in the US?”. Questions can be edited per project.

2. Engines and countries

• ChatGPT and Gemini: answers as shown to users, located in the target country.
• Perplexity and Claude: model answers with web search, located in the target country.
• Google: organic ranking of your domain and the AI Overview for the keyword.

Data is collected through DataForSEO. Answers obtained this way can differ from what an individual user sees in an app, because AI answers depend on time, account and conversation. We therefore repeat questions and report rates rather than single answers.

3. What we extract from each answer

• Whether your brand is named (including known aliases and your domain).
• Its position in the list of recommendations.
• Which competitors are named and where.
• Which sources are cited, with URL and domain, and whether your own site is among them.

4. Metrics

• Mention rate: Share of AI answers to buyer questions that name your brand.
• Average position: Where your brand appears when it is named – 1 means first.
• Share of voice: Your mentions compared with all brands named, including competitors.
• Citation share: Share of cited sources that point to your own domain.

5. Reproducibility

Every raw answer is stored with the parsed result. In the dashboard each cell of the keyword × engine matrix links to the individual answers and to the raw data behind them. The cost of every run is logged.

6. Limits

AI answers vary between runs. A single answer is an anecdote; trends across repeated runs are the signal. zitaso does not scrape consumer apps with fake accounts.