AI visibility

How do I track my brand's AI share of voice in ChatGPT?

Fix a prompt set, run each prompt three times on each assistant logged out, and record cited with a link, mentioned without a link or absent, plus competitors and any wrong facts. Share of voice is your mentions divided by all lender mentions. Start with about 100 prompts and repeat on days 0, 28, 56 and 90.

  • Updated 19 November 2026
  • By Ayan Chakraborty

No cost, no obligation - a written audit within 5 working days.

By Ayan Chakraborty, Founder, SBA Innovation Pvt Ltd, Kolkata. Last updated . Reviewed by Ayan Chakraborty.

How do I track my brand's AI share of voice in ChatGPT?

Answers vary between runs, so a single run proves nothing. Build a fixed set of prompts, run each three times on each assistant, logged out and with search switched on where it is optional, and record the same fields every time: cited with a link, mentioned without a link, or absent; which competitors appear; and whether any rate, fee or rule about you is wrong.

Count a prompt as cited when you are linked in at least two of three runs on that assistant. Share of voice is your mentions divided by all lender mentions across the set, with the competitor list fixed at the start. Run it on days 0, 28, 56 and 90, then monthly, and never change the set or rubric mid-way, or the trend means nothing. Day zero must happen before any change ships. A spreadsheet is enough to start.

  1. Freeze the prompt set, competitor list and scoring rubric.
  2. Run each prompt three times per assistant, logged out.
  3. Log cited with link, mentioned, absent, competitors and wrong facts.
  4. Count a prompt as cited at two or more linked runs out of three.
  5. Compute share of voice against the fixed competitor list.
  6. Repeat on days 0, 28, 56 and 90, then monthly.

How many prompts should a lender track for AI visibility?

Start with about 100 prompts and grow only if you can keep running them consistently. In our consumer lending engagement we used 300 prompts across rates, eligibility, documents, safety and RBI registration and head-to-head comparisons, which gave enough coverage to see movement by prompt group. A smaller set is fine if it covers each intent cluster and you run it the same way every time.

Group prompts by intent so you can see where you win and lose: rules and compliance, calculators and metrics, product questions, trust and registration, comparisons and brand queries. Use real borrower wording, mix English and Hindi where your audience does, and include your own brand questions. The sample below is illustrative; replace the bracketed items with your own lender and products.

A sample borrower-intent prompt set (replace the brackets with your own)
ClusterExample promptWhat it tests
Trust and registrationIs [Lender] RBI registered and safe?Entity facts and registration
EligibilityWho is eligible for a [Lender] personal loan?Extractable eligibility answer
DocumentsWhat documents do I need for a [Lender] loan?Plain-HTML document list
Fees and ratesWhat is the processing fee on a [Lender] loan?Accuracy of fees and rates
ComparisonWhich personal loan lender should I choose in India?Recommendation and shortlist
BrandWhat is [Lender]?Entity clarity

From our client work: the method behind the figures

The consumer lending engagement ran 300 borrower prompts across five assistants each month, three times per prompt, and recorded the median result. The prompt set and scoring rubric were shared in full with the client because they contain only the questions borrowers ask.

Method: Engagement 031: neutral India-located session with no chat history; model updates were not controlled for, which is a real limit on precision, so the prompt set, competitor list and rubric were frozen at the start.

What did not work: Model updates inside the period are not controlled for and limit precision; we cannot separate their effect from our own changes.

Read case study 031 Prompt set, baseline, assumptions and footnotes.

Primary sources

FAQ

Frequently asked questions

Can I automate the prompt runs?

Yes, with tools that support your prompt set and assistants, or with a spreadsheet and manual runs to begin with. Whatever you use, keep the protocol identical so trends are comparable.

Should I run prompts logged in?

Run them logged out, with no chat history, from a consistent location, because personalisation changes answers. Switch search on where it is optional, since that is the mode that cites pages.

What counts as a mention versus a citation?

A mention is the brand named in the answer; a citation is a clickable link to your own page. We count them separately because only citations create measurable sessions.

When do I see the first change?

Perplexity and niche prompts can move in weeks; ChatGPT and Google-based assistants usually take longer. Treat the day-28 run as a first read, and revise targets after it.

Next step

Want this checked against your own campaigns?

The free Growth Audit reviews your funnel, tracking and compliance gaps and returns a written plan within 5 working days.

  • Free of charge
  • Delivered in 5 working days
  • No lock-in