Categories
App Design
Client
Project Overview
This project was created as part of my AI-Driven SEO Systems initiative, where I set out to build an automated internal linking engine powered by embeddings and semantic similarity scoring.
The goal was simple:
Strengthen topical authority, improve crawl paths, and surface underperforming pages faster automatically.
Most websites accumulate internal linking issues over months or years. This system identified them in real time and produced AI-ready linking recommendations that materially improved discoverability, rankings, and session depth.
Categories
App Design
Client
Workflow Architecture
Phase 1 – Data Collection
Crawled 124+ URLs using Screaming Frog (JS rendering enabled).
Exported LLM embeddings for each page to capture semantic meaning beyond keywords.
Pulled engagement metrics from GA4 (engaged sessions, scroll, key events).
Pulled GSC performance data to align linking recommendations with impressions + CTR.
Phase 2 – Data Processing
Cleaned URLs, removed redirects, broken pages, parameter duplicates.
Combined all data sources into a unified dataframe with:
“URL”, “Topic”, “Cluster ID”, “Engagement”, “Semantic Score”, “CTR”, “Keyword Theme”.Reduced dimensionality with UMAP to map pages into semantic space.
Clustered using HDBSCAN to group closely related content.
Phase 3 – AI Analysis
Used embeddings similarity to calculate a “Topical Match Score” for every page pair.
Identified:
orphan pages
pages with weak internal support
strong pages missing contextual links
Used custom thresholds to recommend ideal linking opportunities per URL.
Generated a “Top 5 Recommended Internal Links” list for every page.
Phase 4 – Actionable Output
Produced CSV and visual dashboard showing:
Cluster maps
Linking deficits
Priority pages
Semantic neighbors
Auto-generated internal linking suggestions:
“Page A → Page B” with anchor text ideas.Synced all recommendations to Google Sheets for client visibility.
Real-World SEO Impact
From Embeddings to Rankings:
Boosted organic entrances to under-linked pages by 34%.
Increased crawl-to-index ratio (improving discoverability of new content).
Strengthened semantic depth for competitive service terms (e.g., “furnace tune-up”, “carpet cleaning cost”).
Helped elevate multiple pages from page 3 → page 1 simply by fixing linking gaps.
AI Components Used
Screaming Frog Embeddings Export – page-level semantic vectors
UMAP + HDBSCAN – cluster grouping
GA4 API – engagement metrics
GSC API – CTR & query impressions
Claude / OpenAI – anchor suggestions & cluster labeling
Looker Studio – linking dashboard visualization
Future Enhancements
Integrate weekly auto-alerts for declining pages.
Add GBP review sentiment to strengthen local interlinking.
Deploy AI-generated contextual anchors for each link.
Build a client-facing “Internal Linking Health Score” dashboard.
Branding Nice Studio
I will give you a complete account of the system, and expound the actual teachings of the great explorer of the truth, the master-builder of human happiness. No one rejects, dislikes, or avoids pleasure itself, because it is pleasure, but because those who do not know how to pursue pleasure rationally encounter consequences that are extremely painful. Nor again is there anyone who loves or pursues or desires to obtain pain of itself, because it is pain, but because those who do not know how to pursue pleasure rationally
Categories
Client
Let’s Build What’s Next
From technical SEO automation to deep embedding workflows, I help brands build systems that improve rankings and drive measurable business outcomes.
If you’re exploring AI-powered SEO infrastructure,
let’s make it happen.
