A URL — your SaaS, your store, any public site (or the competitor that keeps getting recommended)
GEO — Generative Engine Optimization
Is ChatGPT recommending you — or your competitor?
GEO Checker scans any URL and shows how ChatGPT, Claude, and Perplexity actually describe it — scored 0–100, with the exact fixes that move the number.
You're on the list. We'll email you at launch.
First access in signup order · founding price locked for your first year
No spam. One email at launch, one if we kill it. That's it.
At a glance
Problem
Buyers now ask AI assistants what to use. Most sites never check what those assistants say about them — or whether they appear at all.
Input
Output
GEO score (0–100) · per-assistant answers · competitor mentions · a prioritized fix list
Time to value
Under 3 minutes
Price
Free: 3 checks / month
Pro: $19/mo
What we check
Twelve checks across three pillars. Every deduction links to a concrete fix — nothing is "just a score".
Visibility40 pts
Are you in the answer at all?
- Category presence
- Whether each assistant names you for buyer-style questions ("best {category} tools", "how do I {job-to-be-done}")
- Share of voice
- Which competitors are named instead of you, and how often, across the query panel
- Query coverage
- Which question phrasings surface you and which never do
- Citations
- Which of your pages assistants actually cite — and whether it's the page you'd want
Accuracy30 pts
Is what they say correct?
- Description
- Does the assistant's one-line summary of you match reality?
- Pricing
- Stale or invented pricing — the most common and most costly hallucination
- Feature claims
- Features attributed to you that you don't have, and ones you have that get missed
- Freshness
- Whether answers reflect your current product or a years-old snapshot
Foundations30 pts
The fixable on-site signals.
- Structured data
- JSON-LD presence + validity (SoftwareApplication / Product / FAQPage), and whether it matches your visible copy
- llms.txt
- Presence and quality of the llms.txt summary AI crawlers read first
- Crawler access
- robots.txt rules for GPTBot, ClaudeBot, PerplexityBot — many sites block them by accident
- Answerability
- FAQ coverage and fact density: can a model quote your pricing, limits, and use cases without guessing?
What a report looks like
Sample output — illustrative, not a real customer's data.
URL example-saas.com GEO SCORE 38 / 100 ChatGPT ✗ not mentioned for "best invoicing tools for freelancers" → recommends FreshBooks, Wave, Bonsai Claude ✗ pricing wrong — says "$49/mo", actual: $19/mo Perplexity ✓ mentioned, but cites your 2023 blog post, not /pricing TOP FIXES (by predicted score impact) 01 +9 pts Add SoftwareApplication JSON-LD with current pricing 02 +7 pts Publish llms.txt — crawlers currently read your nav menu 03 +6 pts robots.txt blocks PerplexityBot (Disallow: /) 04 +5 pts No FAQ answers "how is this different from FreshBooks"
How it works
-
Paste a URL
We read your public pages the way AI crawlers do, then ask ChatGPT, Claude, and Perplexity the questions a real buyer would ask about your product and category — live, not from an index.
-
Read the scorecard
See exactly how each assistant describes you, which alternatives it pushes instead, what it gets wrong, and which of your pages it cites.
-
Ship the fixes
A prioritized checklist — schema markup, llms.txt, FAQ gaps, crawler access — each with its predicted score impact. Re-run after each change and watch the number move.
Pricing
Start free. Upgrade when you want monitoring and competitors tracked.
Free
$0 /mo
- 3 checks / month
- ChatGPT · Claude · Perplexity
- Top 3 fixes per report
- Weekly monitoring
- Competitor tracking
- Score history
FOUNDING PRICE — LOCKED 1 YEAR
Pro
$19 /mo
- 30 checks / month
- ChatGPT · Claude · Perplexity (+ Gemini support coming next)
- Full fix list with predicted score impact
- Weekly re-check + email alert when an answer about you changes
- Track 3 competitor URLs side-by-side
- Score over time, exportable
Questions
What is GEO (Generative Engine Optimization)?
GEO is the practice of making your site visible and accurately represented inside AI answers from tools like ChatGPT, Claude, and Perplexity. It is to AI assistants what SEO is to Google: if SEO got you ranked, GEO gets you recommended.
How is the GEO score calculated?
Three weighted pillars: Visibility (40 points) — whether assistants mention you for buyer-style questions in your category; Accuracy (30 points) — whether what they say about your product, pricing, and features is correct; Foundations (30 points) — the fixable on-site signals: JSON-LD, llms.txt, FAQ coverage, AI-crawler access, fact density. Every deduction links to a specific fix.
AI answers change between runs — is the score even reproducible?
They do vary, which is why a single answer means nothing. GEO Checker runs a fixed panel of buyer-style queries for your category, multiple times per assistant, and scores the aggregate — so one lucky answer doesn't move the number. Re-checks use the same panel, which is what makes before/after comparisons meaningful.
How is this different from an SEO audit?
SEO tools audit how Google ranks your pages. GEO Checker tests what AI assistants actually say when asked about your product and category, then shows the specific gaps — schema markup, llms.txt, FAQ coverage, citations — that cause it.
Which AI assistants does it test?
ChatGPT, Claude, and Perplexity at launch, with Gemini support coming next. Each check runs live queries, so you see what the assistant says today — not a stale index.
Do I need to give you access to my site?
No. GEO Checker reads your public pages the same way AI crawlers do and runs live queries against the assistants. No tracking script, no DNS change, no login. You can also check any competitor's URL.
Does it work for non-English sites?
English and Japanese at launch. Buyer-style queries are run in the language of the site (and in English for Japanese sites that sell internationally), because assistants answer differently per language.
When does it launch?
Early access ships within a few weeks of this waitlist validating demand. Signups are invited in order, and founding pricing is locked for your first year. If we decide not to build it, you get one honest email saying so.