How to track your brand visibility in Claude and ChatGPT (2026)

Overview

Track your brand visibility in Claude and ChatGPT: your AI agent asks the questions your buyers ask in claude.ai and chatgpt.com, and reports whether each answer names or cites your brand, the searches the assistant typed, where your pages ranked in the results of each search, and which pages won instead. For founders and marketers who want Claude and ChatGPT to recommend them when a buyer asks.

Why it's hard

The answer a buyer gets is only on the page of the app. The claude.ai and chatgpt.com apps have no API that returns it, with the searches the assistant made and the pages it cited. An API call to the model is another setup, with other tools and other settings.

The answer also does not tell you why you are absent. An assistant can only cite a page that its own searches returned, and it types its own queries, not your buyer's question. ChatGPT sends several queries at once and gets back one list of results for all of them, without saying which query returned which page. Claude sets a mode on each search, standard or extended, and the two can return different pages for the same query.

And the answer changes. The same question can search today and answer from memory tomorrow, and cite another page each time.

How Reduck does it

Reduck's scripts for claude.ai and chatgpt.com run in your own Chrome, through the Reduck extension, where you are signed in to both apps. Each one opens a new chat, asks one question and waits for the full answer. You need no API key, and each question runs on your own Claude and ChatGPT subscription.

The Claude script returns the answer, which model answered, each web search Claude made with its mode and its results in the order Claude saw them, every page its tools found, and the pages the answer cited. The ChatGPT script returns the answer, the pages it cited, the queries it searched for, and the results of each round, with which ones ChatGPT gave to the model.

Your agent then finds your pages in those lists and tells you, for each question: did the assistant search, which queries it typed, your rank on each, and whether the answer cited you or named you. It reads Claude's queries on Brave too, where Claude's extended search finds most of its pages. The result is a list of the exact queries you lose and the pages that win them, which is what you change your pages for.

Ask the same questions every week, or every day, and compare. To keep the history outside the chat, run the same scripts from the Reduck CLI or the REST API and store each result where you want.

What your agent does

1. Ask each of your buyers' questions in claude.ai and in chatgpt.com, in a new chat
2. Record the answer, the searches the assistant typed and the results of each one
3. Find your pages in the results and give your rank on each search
4. Check whether the answer cited your page or named your brand
5. Read Claude's searches on Brave, with the pages that rank above yours
6. Report the searches you lose, the pages that win them, and what changed since the last run

How to start

1. Create a free account at reduck.ai.
2. Install the Reduck extension in Chrome and pair it.
3. Connect Reduck to your agent: docs.reduck.ai.
4. Ask: "Ask Claude and ChatGPT each of these questions my buyers ask {questions}. For each answer, tell me if it names or cites {my brand}, list the searches the assistant typed, my rank in the results of each search, and the pages that ranked above mine. Then read the Claude searches on Brave too."

Sign in to claude.ai and chatgpt.com in that Chrome first, and turn memory off in both accounts.

AI visibility trackers or Reduck

From each tool's own pages, read on November 2, 2026. "Not documented" means we did not find it on their site.