geoscan.seo7.es is the studio’s free tool. You enter a website address and ten seconds later you see whether an AI can find your page, understand it and cite it in an answer. The check runs across 305 signals in seven categories, and on top of that the tool puts a live question to a language model and shows whose name it gives: yours or a competitor’s.
Client: SEO7’s own project · Site: https://geoscan.seo7.es · Category: GEO / AI tool
The short version
- 305 signals in seven categories weighted by impact, a score from 0 to 100 and a grade from A to F.
- The Reality axis: a live query to the model about your niche and the list of competitors it names instead of you.
- The on-screen result is free, the full report with a PDF arrives by email.
- Three locales: Spanish, Russian, English. Results are written in each language, not machine translated.
- A clean hundred is reachable: irrelevant checks do not penalise the site.
- The lead lands straight in the studio CRM with the website, the score and the competitor list.
Tech stack
- Front end: a single page in plain HTML, CSS and vanilla JavaScript, no frameworks.
- Orchestration: self-hosted n8n in Docker with two workflows: a synchronous scan and the report with the lead.
- Language model: DeepSeek identifies the niche and resolves the Reality axis.
- PDF: a separate Node service with pdfmake and Cyrillic support under systemd, assembling seven category tables, the matrix and the Reality block.
- Infrastructure: nginx with TLS via certbot, security headers, gzip and caching, on the same VPS as the studio’s main backend.
- Localisation: three languages with results written separately for each locale.
The scan is synchronous, one pass, with no Redis and no queues. URL parsing is done with regular expressions: the n8n sandbox has no global URL.
How it started
People stopped googling and started asking. A query like “recommend a dog groomer nearby” or “a good garage in Valencia” returns a ready answer with company names, and your business is either in that answer or it is not.
A site owner cannot check this alone. Ordinary SEO tools measure search positions, not whether ChatGPT reads your site and whether the model has enough data to recommend you.
So we built a tool that answers that question in ten seconds, for free. It measures two things at once: how ready the page is to be read by neural networks, and whether the brand actually shows up in the model’s answer. The first can be fixed with work; the second shows how much work is left.
Full list of what was built
🌐 Interface and analysis
- A check by a single address, with no sign-up and no installation
- A canvas hero visualisation: 305 check names in seven clusters with sparks of connection
- Rotating AI names in the headline
- Three locales with the result in the user’s language
- A cookie notice, localised links, responsive design with its own mobile layout
🧠 Scoring engine
- 305 signals in seven categories weighted by impact
- A GEO Readiness Score from 0 to 100 and a letter grade from A to F
- Core and non-core scoring: irrelevant checks do not penalise, a clean hundred is reachable
- A two-by-two matrix: technical readiness against real citation
🤖 The Reality axis
- A live query to the language model about the site’s niche
- A check of whether the model names your brand or your competitors
- The specific competing domains returned
📄 The report
- Score, category breakdown and matrix on screen, free
- The full report with every signal and a PDF by email
- Inside the PDF: seven category tables, the matrix and the Reality block
🔗 Integrations
- The lead goes straight into the studio CRM with the website, the score and the competitors
- Telegram and email notifications for every lead
- A methodology repository on GitHub: the method described without source code, CC BY 4.0
🛠 Infrastructure and deployment
- A synchronous single-pass scan, no Redis and no queues
- URL parsing with regular expressions
- nginx with TLS, security headers, gzip and asset versioning
Results
| Metric | Value |
|---|---|
| Analysis signals | 305 across 7 categories |
| Check duration | about 10 seconds |
| geoscan.seo7.es own report | 100 out of 100, grade A |
| Interface languages | Spanish, Russian, English |
| Cost to the user | €0 |
| Where the request lands | straight into the studio CRM |
| Openness of the method | GitHub repository, CC BY 4.0 |