1. Sign in and choose questions
Start with saved candidate questions and edit or select the ones you want to observe. Choose a website only when needed. To prepare more questions, select a Search Console property your Google account can access. The initial read preserves separate current and previous 28-day windows, with up to 2,500 returned rows per requested dimension and window. An auxiliary page review reads up to 10 public pages; question preparation uses the saved GSC queries. Select the country, query language, question count, keyword type and sorting priority. getaudit AI turns the selected queries into recommended questions for that region and language; review and edit the resulting candidates. These are questions for you to test, not a record of questions asked on AI services.
2. Choose AI and coverage
Select from the AI platforms available in your workspace, and choose target countries or regions and answer languages. These settings provide question context; they do not establish the network location of the observation.
3. Confirm and review results
Review the brand, exact question wording, AI platforms, models, countries or regions, answer languages, repetitions and total usage. Confirm to start the observation. This saves the chosen question version and settings with their source queries, dates and filters.
Read mentions and citation pages
Each completed record keeps the answer text, returned citation URLs and titles, AI platform, model, collection time and original evidence. The brand flag uses a case-insensitive text match. Read the answer and links together to check which competitors and pages appear.
Keep and export the evidence
Open results by question and export the saved collection as JSON. Confirmed questions, settings and original response hashes stay attached to the run. Failed or unfinished items retain their own status.
Reading the collected data
- A brand text match records a mention; a citation link records a source page. Read each in its answer context.
- Each result belongs to its recorded AI platform, model, question and time. Compare like-for-like runs.
- Use the answer and source links to verify competitor identities; a cited website may also be a publisher, directory or other source.