SEO Content Infrastructure

Overview

An editorial team was running a manual queue on top of Inoreader: duplicates bleeding through, paid SEO credits burning on articles that would never publish, and a backlog past 300 articles with no triage by search value.

The challenge

The editorial pipeline leaked money and time at every stage.

1. Duplicate articles flowed into the queue unchecked.
2. Paid keyword lookups fired on articles that were never going to be approved.
3. Basic AI prompts returned keyword slop instead of the real terms in the article body.

The approach

1. Ingests Inoreader feeds every 12 hours with URL-level dedup, full-body fetch, and AI classification for audience, content type, and a 1 to 10 quality score.
2. An approval gate holds everything: operator approval is the only trigger that releases an article into paid enrichment.
3. Keywords are extracted from the article body itself and checked against a keyword master, so nothing is ever paid for twice.
4. The outline generator pulls only sources scoring 5+, then returns a publish-ready H1 to H3 outline in under 10 minutes.

The results

1. Four-stage editorial state machine governs every article from Pending to publishable outline
2. Approval gate blocks paid SEO enrichment on rejected articles, collapsing credit burn to near zero
3. Keyword dedup layer reuses prior enrichment across campaigns, so cost compounds downward
4. 1 to 10 AI scoring filters weak sources before outline generation
5. Delivered with an operator guide, so the editorial team runs the pipeline themselves