Stratum crawls your site and the top 30 SERP results, builds knowledge graphs of both, then scores your content against six gap types — telling you precisely what to write and why.
Every other topical gap tool measures your site against what currently ranks. Stratum goes further — measuring against what's actually knowable. That's the difference between chasing competitors and understanding the full opportunity.
Built by crawling your domain. Maps every entity, predicate, and relationship claim your content currently makes — what you say, how you say it, and how it connects.
Built from the top 30 SERP results for your target query. Establishes the reference standard — what the content that currently ranks actually claims and covers.
Built from primary knowledge sources — Wikipedia and Wikidata SPARQL. Measures your coverage against what's actually known, independent of what competitors cover.
Stratum runs locally in your browser — start a Python server, open localhost, and your data never leaves your machine.
Paste your target domain or sitemap URLs. Stratum auto-discovers sitemap structure via robots.txt — Playwright handles JS-rendered or Cloudflare-protected pages automatically.
Stratum previews the discovered URL set. You review, set max pages, and confirm before the crawl begins. Full control over scope before any processing runs.
Concurrent Playwright crawl (5 workers default) runs TF-IDF, LSI, spaCy NER, LLM entity extraction, and Neo4j graph writing per page. Real-time SSE progress throughout.
Enter your target query. Stratum scrapes the top 30 results, builds G_D, and projects your site into SERP embedding space — generating the 3D competitive landscape.
A prioritised, gap-typed URL table tells you which pages need work and why. Six gap types — predicate absence weighted highest because fixing a relationship claim is lower effort than creating a new page.
The 4-step Content Writer uses SERP intelligence — consensus entities (cross-SERP frequency-weighted) and information gain opportunities (what competitors collectively miss) — to generate targeted briefs and replacement content.
Stratum's architecture borrows from knowledge representation, epistemic logic, and information theory — disciplines that don't usually touch SEO tooling. The result is gap analysis that goes significantly deeper than keyword coverage.
Predicate absence — missing relationship claims between entities — is weighted highest (β=0.30) because fixing it requires editing, not new page creation. Entity absence is next. Neighbourhood incompleteness, depth, authority, and query-response gaps complete the picture.
Every Subject-Predicate-Object triple extracted from your content is assigned an epistemic tier — Factual (>0.8 belief), Assumption (0.5–0.8), or Derivative (<0.5). No other SEO tool distinguishes between what content asserts as fact versus what it implies or derives.
Every node and edge in the knowledge graph carries valid_from and valid_until timestamps. A product specification claim has a different temporal stability class than market data. Stratum tracks seven stability classes — from permanent metadata to market data — because not all content claims age the same way.
A Three.js interactive map plots 30 SERP results across depth layers and topic clusters. Your page is projected into SERP embedding space so you can see exactly where you are relative to what ranks — and the Content Writer generates a brief from the gap directly.
The Content Writer's intelligence panel shows consensus entities (cross-SERP frequency-weighted, with coverage badges) and information gain opportunities — topics competitors collectively miss. Briefs and generated content reflect both what you must cover and what you alone could own.
Stratum runs as a local Flask server — Neo4j and Postgres run on your machine in server mode. No cloud database, no telemetry, no data leaving your environment. The only external calls are LLM API requests using your own keys and Playwright SERP scraping.
The data pipeline, crawl system, and gap scoring framework are built. The knowledge graph intelligence layer — entity resolution, trust scoring, and information gain — is next. Register and we'll keep you updated as each capability ships.
