GuidesAI search

Tutorial: tracking your brand's visibility in AI answers

By the Banzai Labs teamUpdated on 2 min read

QuestionCitedCitation rate, per monthThe same questions, asked every month
A fixed list of questions, asked to AI assistants every month, and the brand’s citation rate.

1. The question list

Thirty to fifty real questions from your customers, in three families:

Family Example
Generic “Which agency should build my mobile app?”
Comparison “Shopify or WooCommerce for a small store?”
Brand “Is [your brand] any good?”

2. The record

date,assistant,question,brand_mentioned,link_to_site,position,competitors_cited
2026-11,ChatGPT,"Which agency should…",yes,yes,2,"A;B"
2026-11,Perplexity,"Which agency should…",no,no,,"A;C"

3. The metrics

  • Mention rateShare of questions where the brand appears.
  • Citation rateShare of answers with a link to your site.
  • CompetitorsWho is cited instead of you, and on which questions.

4. Cross-check with traffic

Visits from assistants show in GA4, and their bots’ visits in the server logs. The three measures together tell you whether your efforts are paying off.

Traffic: measuring ChatGPT traffic in GA4. Bots: spotting AI bots in your logs. To get cited: generative engine optimization.

Frequently asked questions

Do answers change from one time to the next?
Yes, even for the same question. That's why you track a rate across dozens of questions rather than a single answer.
Are there tools for this?
Several services automate the check. A spreadsheet and an hour a month are enough to start and understand what you're measuring.
Does the signed-in account affect the answer?
It can: history and memory personalise answers. Use a session with no history for the check.

Read next

All guides