Case Study

I built a free tool that shows a business how ChatGPT, Gemini and Claude actually talk about them: real buyer questions, live web search, and a score with receipts behind every number.

The GEO Audit landing page: see how ChatGPT, Gemini and Claude talk about your business

Expand

GEO Audit Tool

Generative engine optimization, measured · run a free audit · built June 2026

Product Design Web App Build Live AI Search Plain-English Reports PDF & Email Delivery

What it does

You enter a business. It asks ChatGPT, Gemini and Claude what a real customer would ask, then grades every answer for whether the business came up and where it ranked.
Every score has receipts: open any question to read the engine’s actual answer and see which websites it pulled from.
You get the fastest fixes in plain English, quick wins first, plus a 4-page PDF to download and a copy emailed to you.

AI engines queried live on every audit

real buyer answers read and graded per audit

~60s

from a URL to a report, PDF and email included

The Challenge

People stopped Googling. Businesses had no way to see what replaced it.

Buying decisions now start with a question to ChatGPT instead of a search bar, and no owner has any idea what it says about them. That is what GEO means: generative engine optimization. Same game as SEO, except the engine answers in one sentence instead of a page of links, and if you are not in that sentence you do not exist. No SEO tool measures it, so I built the one that does.

The Solution

One audit, three layers.

SYSTEM 01

The audit engine

It works out what kind of business you are, picks five questions a real customer would ask, and sends them to all three engines with live web search on. A grader reads every answer: did your name come up, and where. That plus a check of how readable your site is to AI becomes the score.

I ran it on one of the most famous pizza spots in New York. Same question, same minute: Gemini put them at #1, Claude at #2, ChatGPT never mentioned them at all. One blended “AI score” would have buried the only thing worth knowing.

on Gemini

on Claude

mentions on ChatGPT

Per-engine results: ChatGPT 0 percent, Gemini 60 percent, Claude 60 percent — Three engines, one business, three different answers

The score card: 45 out of 100 with visibility and AI-readiness sub-scores — The score it rolls up to

Python · OpenAI · Gemini · Claude

SYSTEM 02

Source citations behind every score

Every topic opens into the questions asked, and every question opens into the engine’s real answer with a verdict on it. The citations section shows which websites each engine read to get there, and flags when one of them is you. Nobody has to trust the score.

Topic drill-down: the exact questions with per-engine mentioned and not-mentioned badges — One question, three engines, three verdicts

3-level drill-down · Source citations

SYSTEM 03

The fix list, ordered by effort

Every fix is written for a business owner, not an SEO. No jargon. And the fixes come ranked: quick wins first, bigger projects after. Each one names the exact gap, which is how you go from invisible on an engine to showing up. If your site was built with Claude or any other AI, the whole list pastes straight back in as instructions and most of it gets fixed in one sitting.