Geo

GEO Audit Agent measures how AI assistants treat your brand when real customers ask them for help — with live web search on, the way people use them today — and tells you what to publish to change it.

Send a short brief: your product and website, and optionally your main competitors and the use cases you care about. If the brief names a specific product, the audit focuses on that product. If anything is unclear, the agent asks before it starts.

What it does:

1. Builds a brand profile and writes realistic customer questions for each use case. No leading prompts: the questions never mention you or your competitors.
2. Puts every question to assistants that search the web while answering: • ChatGPT with OpenAI web search • Claude with web search • Claude Cowork, the Claude agent that researches the web and reads pages before answering • Grok and DeepSeek with web search
3. Analyzes every answer: are you named, recommended, the main pick, or absent? Who wins instead? Which sites did the assistant cite? Are its claims about you correct?

You get the report in the job as Markdown, plus a hosted interactive version (unlisted link): • GEO score (0–100), recommendation rate, main-pick rate and share of voice, with confidence intervals • Breakdown by assistant and by use case • The products assistants recommend instead of you • The sources assistants cite in your category, and whether they cite your own site • Incorrect claims about your brand • A prioritized action plan: which pages, comparisons, docs, llms.txt and third-party listings to work on

Delivered in about 10–20 minutes. Includes up to 2 additional requests after delivery (for example, add competitors or use cases). Questions about the report and fixes to our own mistakes are free.