Monitor what AI says about our brand

What it is

Brand monitoring now includes AI answers, alongside search results, reviews and social. The goal is to know whether AI tools mention you for the questions buyers ask, what they say about you, and which sources they lean on.

How it works

1. Write a fixed prompt set covering category, comparison and brand questions, with several phrasings for each. Size it to what you can review; IonBin doesn’t set a number.
2. Run each prompt repeatedly on a schedule across the engines your buyers use, and report visibility as the share of runs that mention you, not as a rank.
3. Log mentions, sentiment, factual errors and cited domains.
4. Find content gaps. Rand Fishkin describes exporting your brand mentions and asking an AI model to compare them with your site, to surface questions and doubts your content doesn’t answer (view: Fishkin; the tool he names, Alertmouse, is his own product).
5. Fix errors at the source. Aleyda Solís suggests finding the pages behind a wrong answer, correcting the fact across your own and third-party footprint rather than only the cited page, then rerunning the same prompts (view: Solís). See when AI gets your brand wrong.

Benefits

• Early warning on wrong facts.
• Visibility into which sources shape answers.
• A baseline for AEO work.

Limits and risks

• Single checks are unreliable. In a SparkToro and Gumshoe study published 2026-01-28, 600 volunteers ran 12 prompts 2,961 times across ChatGPT, Claude and Google’s AI features. Fewer than 1 in 100 runs returned the same list of brands, and about 1 in 1,000 the same order (SparkToro). Fishkin’s view is that share of runs mentioning you may be a reasonable metric, but the choice of prompts is a large weakness.
• The sources behind answers are opaque. Simon Willison points out that Anthropic and OpenAI lean heavily on search yet don’t disclose the search index they use (view: Willison).
• Automation is restricted (terms checked 2026-10-17). OpenAI’s terms for ChatGPT prohibit automatically or programmatically extracting output. Anthropic’s consumer terms bar accessing Claude by bot or script except through an Anthropic API key. Perplexity’s terms bar using bots, scrapers or automated queries to extract data from its services. Google’s terms bar automated access that ignores its machine-readable instructions, such as robots.txt. Use each vendor’s API, or a tool that does, rather than scripting the consumer apps.

This page summarizes published rules for orientation. It isn’t legal advice; check the rules that apply where you and your recipients are.