What does ChatGPT say about my brand?

How do I find out what ChatGPT says about my brand?

You can check by hand today. It takes about an hour and tells you whether you have a problem worth tracking.

1. List the questions people actually ask. For example: "What is [brand]?", "Is [product] any good?", "Best alternatives to [product]?", "Is [founder] trustworthy?", "Should I buy from [brand]?"
2. Ask ChatGPT each question. Run it once with web search on and, where you can, once with search off. The two answers can differ a lot, and the difference matters (see below).
3. Copy the answers verbatim into a spreadsheet, with the date and any sources ChatGPT cited.
4. Score each answer positive, neutral, or negative, and note the recurring claims, both the flattering ones and the complaints.
5. Repeat the same questions monthly and compare. The change between rounds tells you more than any single snapshot.

Why does ChatGPT say different things about my brand on different days?

ChatGPT can answer in two ways, and the wording changes depending on which one it uses.

From training data

When ChatGPT does not search, it answers from what it learned during training. That knowledge has a cutoff date, often months in the past, so it can repeat an old review or a 2023 forum complaint as if it were current.

From live web search

When ChatGPT searches first, it reflects what is published about you right now and cites the pages it read. This is closer to what a careful user sees today. Models also get retrained and swap underlying versions, which is another reason the answer drifts. Saidly runs every model with live web search on, so reports mirror what real users get.

Why does it matter what ChatGPT says about my company?

When someone wants to know about your company, there is a growing chance they ask an AI assistant before a search engine. The one-paragraph answer they get often decides whether you make the shortlist, and you never see it.

• The answer is private. A bad review sits on a page you can find and respond to. An AI answer is generated for one person and disappears. Without monitoring, you have no idea what was said.
• Stale claims stick. A model trained on old pages can keep repeating a complaint you fixed last year, until you find the source and correct it.
• Each model tells a different story. On the same day, ChatGPT, Claude, Gemini, and Grok can describe you differently because they read different sources.

This is the measurement side of AI visibility monitoring. For the scoring method behind it, see AI brand sentiment.

Manual checking vs. tracking it with Saidly

The manual method is free and worth doing once. It also breaks down quickly: it is a snapshot of a few questions on one day, hand-scoring is subjective and inconsistent between rounds, it covers only one model unless you do everything four times, and most teams stop after the second month.

Pro and Business plans add two lenses: "model-only" (what ChatGPT believes without the web) and "compare" (live web vs model-only, showing the gap). You can also see a sample report before you sign up.

Which AI assistants should I track?

Track all four flagship consumer assistants, because buyers use all of them and each cites different sources:

• ChatGPT (OpenAI), the most widely used assistant.
• Claude (Anthropic).
• Gemini (Google).
• Grok (xAI).

Saidly is built by Woodfire Digital, LLC and covers all four, grounded on the live web, in a single report. Plans start at $2.99/mo with a custom Enterprise tier; the FAQ explains how the search modes work.

See what ChatGPT and the others say about you

Pick your questions, track ChatGPT, Claude, Gemini, and Grok on a schedule, fix the sources, and re-measure. Saidly handles the measurement so you can focus on the fixing.

30-day free trial. No credit card to start. Read the full guide or browse the FAQ.