We run the questions a real buyer asks across ChatGPT, Claude & Gemini — then grade every answer against your own facts. Wrong claims, competitor rankings, missed differentiators — all ranked by severity.
Fabricated product recall
Claimed a 2023 recall that never happened.
Ranked competitor #1
Recommended Acme over you for enterprise cold-chain.
Wrong HQ location
Listed Boston instead of Cambridge across 2 of 3 runs.
Missed key differentiator
None mentioned your FDA 21 CFR Part 11 audit trail.
Graded across the models your buyers actually use
AI models interrogated
ChatGPT, Claude & Gemini — the assistants buyers trust
runs per scan
every question ×3 per model, to control for randomness
dimensions scored
accuracy · visibility · competitive · sentiment
claims fact-sourced
every fix we draft cites your own verified facts
What we catch
Every finding is ranked by severity and tied back to the exact buyer question and model that produced it.
Fabricated recalls, denied certifications, wrong locations, mistaken identities — every claim graded against your real facts, so you catch the answers that create risk.
Whether you even appear when a buyer asks for the best vendors — and exactly where you land relative to the alternatives.
We ask the model who it would pick and why — then catch it quietly deferring to a competitor using strengths you actually have.
Every scan ends with prioritized, copy-paste fixes — for your website and beyond it — with every claim sourced from your verified facts. Make the change, re-scan, watch the score move.
How it works
We under-report before we over-report. Every flag is something you'd actually want to fix.
We read your site, then you confirm the truth set — certifications, products, differentiators — and add your own buyer questions.
A consistent set of real buyer prompts runs across Claude, GPT & Gemini — three runs each to control for non-determinism.
An independent judge grades every claim, ranks what matters, and the fix plan turns it into changes you can ship.
FAQ
The short version of what a scan is, what it costs to start, and where the fixes come from.
It runs the questions a real buyer asks across ChatGPT, Claude and Gemini, grades every answer against your verified facts, and ranks what it finds — wrong claims, competitor positioning, missed differentiators — by severity, ending in a prioritized fix plan.
ChatGPT (OpenAI), Claude (Anthropic) and Gemini (Google). Every buyer question runs three times per model — nine runs per question — to control for the randomness in model answers.
Four dimensions: accuracy (are the claims about you true?), visibility (do you show up at all?), competitive position (where do you land when a buyer asks for the best options?) and sentiment (how you are framed).
No. A scan starts with just your business name and website — no signup required to see results.
Every fix is drafted from your own verified facts — the truth set you confirm at the start of a scan — so each change ships with its source. Fixes are copy-paste ready: llms.txt fact sheets, schema.org structured data, FAQ blocks, robots.txt corrections and more.
SEO optimizes where you rank in search results. GEO (generative engine optimization) is about what AI assistants actually say and cite when a buyer asks them directly. ReputationIQ audits whether AI crawlers can reach you, grades what the models currently claim, and gives you the fixes that change those answers.
a few minutes · no signup to start · reputationiq.ai