Local draft. Highlighted items are placeholders that need the owner’s confirmation before launch.

    Press / to search

    Get started
    Menu

    Why numbers differ from ChatGPT

    Last updated

    AI answers change from one run to the next, so the answer you get in your own chat is one sample, while Aivting's figures count many collected answers.

    Who can do this: everyone.

    Steps to compare fairly

    1. Open the answers behind the number. See View the answers behind a number.
    2. Open one answer and check Prompt sent (exactly as sent to the engine) and Location and language. Your own chat may use different wording, a different location or your chat history.
    3. Check the sample size under each figure, for example "12 of 40 answers name Brindlepath". A figure based on a few answers moves a lot.
    4. Pick a longer date range to base the figure on more answers.

    Why the numbers differ

    • Answers vary. The same prompt can name different brands from one run to the next. Each run collects several answers per prompt from each collected engine, so a single chat is one of many possible answers.
    • Your chat isn't neutral. Your account, memory, location and earlier messages can change what you see.
    • We count only collected answers. "No answer shown" and "Failed" don't count, and never count as "not mentioned".
    • Only tracked brands count. Share of voice and position compare your brand with the competitors you track, not every brand in the answer.
    • Removed prompts don't count. Figures use active prompts only.

    Figures are estimates based on the number of answers shown.

    If it still looks wrong

    Contact us with the prompt, the date and a copy of the answer you saw.

    Still stuck? Contact us.

    ChatGPT is a trademark of OpenAI. Google and AI Overviews are trademarks of Google LLC. Aivting is an independent product and is not affiliated with or endorsed by OpenAI or Google.