2026-09-14 — Brand Monitoring · Fundamentals

How to track what AI assistants say about your brand

People increasingly ask an assistant instead of searching. If your brand isn’t in the answer, or a competitor’s page is the cited source, you won’t see it in your analytics. Here is a setup that measures it.

1. Choose the prompts your buyers ask

Start with 50–200 prompts grouped by intent: category (“best project management tool for agencies”), comparison (“Asana vs Monday”), and problem (“how to track client hours”). Keep the wording natural; assistants answer differently to keyword strings.

2. Run each prompt where your customers are

Answers change by country and even by US state. Run every prompt in each market you sell in, with the country (and state) parameter.

3. Queue the runs as async tasks

Submit the whole set as one batch with an idempotency key per prompt, market and day, so a retry never double-counts:

[
  { "taskType": "CHATGPT", "payload": { "prompt": "best project management tool for agencies", "country": "US" }, "idempotencyKey": "pm-agencies-us-2026-09-14" },
  { "taskType": "CHATGPT", "payload": { "prompt": "best project management tool for agencies", "country": "GB" }, "idempotencyKey": "pm-agencies-gb-2026-09-14" }
]

Add a webhook URL and results arrive as each task finishes.

4. Measure three things

• Mention rate: the share of answers that name your brand (check entities and text).
• Citation share: how often your domain appears in sources, and at which position.
• Fan-out coverage: whether you rank for the searches the assistant ran (searchQueries). Those are the queries to optimise pages for.

5. Re-run on a schedule

Answers drift week to week. A weekly run of the same set is enough to see trends without paying for noise; daily runs make sense for launch weeks and competitive categories.

Keep reading