# Your client just asked what ChatGPT says about them. Here's the framework

Published: 2026-08-06
Author: Brandflare Team (https://brandflare.io)
Canonical: https://brandflare.io/blog/what-does-chatgpt-say-about-your-clients

**The short answer:** don't guess, and don't stall. Ask the engines yourself before the meeting, sort what you find into four buckets — present and accurate, present but wrong, absent, or misattributed — and walk in with the buckets instead of a shrug. The question isn't going away; it's the 2026 version of "why aren't we ranking?"

## Why this question is suddenly everywhere

Your client's customers changed how they shop for answers. When someone asks ChatGPT "best CRM for a 10-person agency," the reply names three or four brands — and if your client isn't one of them, that's a shortlist they never knew they missed. No impression data, no referral string, no consolation-prize ranking on page two. Just absence.

The uncomfortable part for agencies: clients assume you already know what the engines say. You do SEO, this looks adjacent, therefore it's your department now. Congratulations on the promotion.

## The four buckets

Every engine answer about a brand lands in one of four buckets, and each bucket has a different fix. This is the sorting that turns a vague worry into a work plan.

- **Present and accurate.** The engine names the client and gets the facts right. Your job is defense: keep the sources fresh so it stays true.
- **Present but wrong.** Named, but with stale pricing, a dead product line, or a description from two rebrands ago. Usually a [training-data](https://brandflare.io/learn/glossary/training-data) echo — fixable, but through the sources engines read, not through wishing.
- **Absent.** Competitors get named; your client doesn't. Almost always a footprint problem: the comparison posts and review roundups engines lean on simply don't include them.
- **Misattributed.** The engine blends your client with a similarly named company. Entity ambiguity — annoying, common, and very solvable with consistency work.

## Before the meeting: a 30-minute sweep of your own

Ask ChatGPT, Gemini, Perplexity, and Claude the three or four questions the client's actual customers would ask. Not "tell me about Acme" — nobody shops that way. Ask category questions: "best X for Y," "alternatives to [market leader]," "is X worth it." Record who gets named, in what order, and what gets claimed.

Two things to watch: whether answers carry [citations](https://brandflare.io/learn/glossary/citation) (that's live retrieval — you can trace the source) and whether they don't (that's memory — a longer game). The split tells you which fixes are fast and which are quarterly.

## After the meeting: promise measurement, not magic

The honest pitch is that AI answers can be moved — engines ground answers in sources, sources can be improved — but the timeline is weeks for retrieval-driven answers and quarters for memorized ones. Promise a baseline, a monthly re-check, and movement on specific answers. Don't promise "we'll get you into ChatGPT" as if there's a form to fill in. There isn't, and your client will eventually ask to see it.

A [free audit](https://brandflare.io/audit) gives you the baseline in about five minutes per client — all four engines, real customer questions, scored and shareable. Run one before the meeting and you'll be the only agency they've talked to who showed up with data.