Category 01
Structure
Headings, sections, and hierarchy. Does your page read like a clear outline or one long block of marketing copy?
Labs / GEO Scoring Tool
Paste any public URL and see how legible your page is to AI assistants. Get a scored breakdown across six GEO categories, plus targeted fixes.
Run a GEO Score
Paste a public URL to see how AI assistants read your page.
Use Cases
01
Use the test when you ship a new landing page, product page, or thought piece and want to know if AI assistants can restate the offer cleanly before it goes live.
02
Check a few key URLs from your portfolio pages, solution pages, and FAQs to see where AI might be guessing about locations, sectors, or pricing.
03
Use scores as a neutral, data-backed way to explain why you are pushing for better content, schema, and internal links across a client account.
When not to use it
If a page is behind authentication, contains confidential data, or is still a rough draft, do not test it. Score pages that are live and ready to perform.
The shift
Search is moving from ten blue links to a single answer. People ask AI, get one response, and move on. If models struggle to read your page, your brand stays out of that answer.
The GEO Crash Test gives you a directional signal. It shows where your page is easy for assistants to use and where you are forcing them to guess — and it gives you a concrete list of fixes.
You cannot control what a model does with your content, but you can control how clearly you present it.
"If AI cannot restate your offer in one clean paragraph, you are already losing to the teams who can."
— Nicholas PutzScoring
These are the same categories that appear in your result card. Strong scores here make your page easier for assistants to trust and reuse.
Category 01
Headings, sections, and hierarchy. Does your page read like a clear outline or one long block of marketing copy?
Category 02
Who you are, what you do, for whom, and where. Can an assistant answer those basics without making assumptions?
Category 03
Common questions buyers ask. Does this single page contain enough substance to support useful answers?
Category 04
Schema, metadata, FAQs, and internal links. Do you give models clean hooks to pull facts from your content?
Category 05
How well your page compresses into a machine-friendly summary without losing meaning or context.
Category 06
Crawlability, titles, descriptions, and performance signals that affect whether your page is seen at all.
Output
Free tests return your score and one top fix. Pro unlocks the full implementation plan.
Methodology
The scoring categories and general method are open. The exact prompts, weights, and thresholds stay proprietary and will evolve as AI search changes.
The goal is a steady reference point you can watch over time, not a magic number that pretends to be perfect.
Want to see an example result? View a sample GEO Crash Test JSON →
Machine view
Definitions
A GEO score is a numeric rating from 0 to 100 that measures how easily an AI assistant can read, understand, and reuse the content on a specific web page. Higher scores indicate that the page is structured, clear, and complete enough for AI to generate accurate answers using your content as a cited source.
GEO (Generative Engine Optimization) optimizes content for AI assistants like ChatGPT, Perplexity, and Google AI Overviews, while SEO targets traditional search rankings. SEO prioritizes keyword signals and backlinks; GEO prioritizes clarity, factual density, and structural completeness so AI can accurately cite your content when generating answers.
The GEO Crash Test works by fetching your page's HTML, stripping scripts and styling, and sending the cleaned content to an AI model that scores it across six categories: structure, clarity, coverage, schema signals, AI legibility, and technical basics. Results return as a scored card with prioritized recommendations.
The GEO score measures how well a single web page communicates its content to AI assistants. It evaluates six categories: structural hierarchy, identity clarity, topical coverage, schema and metadata signals, AI legibility, and technical crawlability. Each category contributes to the overall score, with the lowest-scoring areas surfaced as priority fixes.
Methodology & Limits
GEO Crash Test uses AI models that work with probabilities, not fixed math. If you run the same page a few times you may see small shifts, usually a couple of points. This is normal noise in how the model reads and compresses your content — not your site suddenly changing. Look for big moves and consistent patterns, not tiny wobbles.
GEO Crash Test combines a static frontend with a proprietary scoring function. The function fetches your page HTML, cleans it, and sends a trimmed version to a model that returns a structured JSON score. The categories, weights, and prompts are custom and evolve as AI search changes. The backend runs on Netlify Functions with OpenAI.
Each run is a snapshot of your page at the moment you test it. Scores change as you update content and as models change. There is no continuous monitoring in the current version.
No. The GEO Crash Test uses representative models and controlled prompts. It is a directional signal, not a mirror of every system in the wild. Use it to understand your baseline and track improvement over time.
It is not. A strong score means your page is easier for AI to understand and reuse. It still competes with other brands, other content, and other signals. Think of it as a necessary condition, not a sufficient one.
The system fetches your page, scores it, and returns the result in your browser. There is no separate database of your scores beyond standard infrastructure logs. If aggregated, anonymized reporting is added in the future, this page will be updated to explain what is collected and how it is used. Your data is not sold.
It is not legal advice, it is not a guarantee of performance, and it does not replace human judgment. It is a clear starting point for better decisions about how your page talks to AI models.
GEO Content Cluster
These articles explain the concepts behind your score — what GEO is, how it differs from SEO, and how AI Overviews decide what to cite.
Weekly newsletter
Every Tuesday: frameworks for AI-enabled marketing, GEO tactics, and brand strategy built to outlast the next technology shift. Written by Nicholas Putz.
Subscribe on SubstackWork together
If your score surfaces gaps you want help fixing — or if you want a full GEO audit across a site or portfolio — PureDigital can build the implementation plan and execute it.
Explore PureDigitalPro access unlocks the full implementation plan, all recommendations with step-by-step tickets, technical notes for your dev team, and the machine view of your page. Payment is handled through Cash App or Venmo. After payment you receive a private Pro access code you enter in the form above.
After sending payment, use the contact form below and include your email, payment method, and the
site you want to test. I will reply with a Pro access code.
Terms: This purchase gives you access to Pro scoring output only. No promise of rankings or traffic.
Sites that block automated requests cannot be fetched — if that happens, choose a different public page.
No refunds.
GEO Crash Test runs on paid model credits and infrastructure. If this tool helps your work and you want to keep it free for others, you can chip in toward the API bill.
Completely optional. The test stays free to use either way.
Press & Research
If you write or speak about AI search, GEO, or the future of discovery, you can reference GEO Crash Test scores in your work.
See another AI-enabled tool built by Nicholas Putz: Lawn Intel — real-time lawn care scheduling using weather data and AI-calibrated timing.
Citation
If you reference GEO Crash Test scores in a report, article, or presentation, here is a pre-formatted citation block.
APA style
Putz, N. (2025). GEO Crash Test [Web application]. PureDigital / nicholasputz.com. https://nicholasputz.com/labs/geo-crash-test/
Plain text / web copy
GEO score via GEO Crash Test by Nicholas Putz (nicholasputz.com/labs/geo-crash-test/), a free tool that measures how well a web page communicates its content to AI assistants.
Linking to https://nicholasputz.com/labs/geo-crash-test/ as the canonical source is appreciated and helps keep the tool free.
For press inquiries or permission to reproduce the scoring methodology, visit the press page.
Contact
If you want help improving your score or applying this thinking across a portfolio, send a note.