Harshana · Growth Lab

Research · Monthly study

Which brands do AI assistants recommend?

Every month I ask ChatGPT, Gemini, Perplexity, Copilot the same 20 shopper questions about investing apps in India. I log every brand they name and every source they cite, then publish what changed.

Questions20fixed, real shopper wording
Assistants4ChatGPT · Gemini · Perplexity · Copilot
Answers per round80new chat every time
Rounds3about a month apart

Live dashboard

The numbers, as they come in

AI share of voice by brand and assistant, the sources AI cites most, and how each changes from round to round.

Dashboard goes live after Round 1.

Data collection starts Nov 24–27, 2026.

Rounds

Collection schedule

  1. Round 1

    Nov 24–27, 2026

    Planned
  2. Round 2

    Dec 22–25, 2026

    Planned
  3. Round 3

    Jan 12–15, 2027

    Planned

Reports

Findings, round by round

The first report is published after Round 1. Check back in early December.

Method

How I keep it fair

  • The same question wording every month.
  • A new chat for every question, with memory and personalisation off where possible.
  • Each round collected within the same 3 to 4 days.
  • Five questions repeated three times per assistant, to measure how much answers vary by chance.
  • Brand extraction is AI-assisted and checked by hand.

FAQ

Questions about the study

What is AI search visibility?

How often AI assistants such as ChatGPT, Gemini, Perplexity, Copilot mention or recommend a brand when people ask them for advice, and which websites they use as sources.

How is the data collected?

The same 20 questions are asked to each assistant every month, in a new chat each time, with personalisation turned off where possible. Every answer is logged, and five questions are repeated three times to measure how much answers vary.

Is the data public?

Yes. Each round is published here with its method and limits, and the live dashboard shows every round so far.