Prompts: questions and answers

Each question is asked 3 times per AI every week, because answers vary between runs. Claude is asked once a week, because each of its answers runs a web search.

We report two rates: the share of answers that mention your brand and the share that cite (link to) your site, each with a 90% confidence range. We never report a “position”: answers change too much between runs for one to mean anything.

Questions and AI tools per plan: Starter: 15 · ChatGPT, Gemini, Google AI Overviews, Google AI Mode, Perplexity; Growth: 30 · ChatGPT, Gemini, Google AI Overviews, Google AI Mode, Perplexity; Business: 60 · ChatGPT, Gemini, Google AI Overviews, Google AI Mode, Perplexity, Claude.

Full detail in the methodology.

Each row is a question and each column an AI. At the top, They link to you in X of Y questions and Competitors win N: the questions where AI recommends competitors and not you. The number next to Prompts in the sidebar counts those with nothing under way yet.

Open a question to see what each AI actually answered, every sample, with the brands and pages it cited.

Click Manage questions to add, pause or remove. Paused questions keep their history; your plan sets how many are measured at once. Questions we suggested when you added the site are marked Suggested.

A good list mixes types: discovery (“best … for …”), comparison, alternatives, problem and price. Keep questions that name your brand to a few: they only tell you whether AI knows you, not whether it recommends you.

Need ideas? Research a topic in Articles suggests the questions buyers ask, with sources.

A change is shown as up or down only when the ranges of the two weeks say so; otherwise it reads no clear change yet. A single week rarely proves anything.

Higher plans measure more questions (Business also adds Claude); Grok is an add-on for any plan, asked once a week in your main country. Languages are measured separately: each question has its own language and country.