SECTION 01 / HERO
Home/Case Studies/TheFeedLab.io
Case Study · RSS Content Intelligence Engine

TheFeedLab.io — Signal
Over Noise, Automatically.

An RSS intelligence engine that scores every content candidate on four dimensions and surfaces only the highest-value items — removing the manual scroll from content operations entirely.

Active Buildn8nRSS AggregationWrite-Score LayerWordPressTelegram
SECTION 02 / OVERVIEW
02 — Overview

Most Content Teams Spend Hours
Finding What to Write About.

TheFeedLab replaces the manual content discovery process with a scored signal layer.

Content operations at scale have a well-documented bottleneck: finding and evaluating what to write about. Manual RSS reading, trend monitoring, and topic scoring are time-intensive and inconsistent. TheFeedLab solves this with a write-score layer that evaluates every content candidate automatically.

The system aggregates RSS feeds from configured sources, applies the write-score algorithm to each item (trend momentum + audience fit + content utility + monetization fit, minus risk penalty), and pushes the highest-scoring candidates to a Telegram trending alert bot. Content teams act on scored signals instead of scrolling through feeds.

The underlying pattern — ingest → score → surface → publish — is the same architecture that retail operators use for review monitoring and freight operators use for load board scanning. TheFeedLab is the content intelligence instantiation of a general signal processing pattern.

Stackn8n RSS aggregation · custom write-score layer · WordPress REST API publishing · Telegram trending alerts · Cloudflare tunnel (dev.thefeedlab.io)
n8nRSS FeedsWrite-Score LayerWordPressTelegram BotCloudflare Tunnel
4
Score Dimensions
24/7
Feed Monitoring
<5m
Alert Latency
0
Manual Checks Required
Write-ScoreTrend momentum + audience fit + content utility + monetization fit − risk penalty
SECTION 03 / PIPELINE
03 — Pipeline

Four Stages. Feed In,
Scored Signal Out.

Stage 01

RSS Aggregation

Polls configured RSS feeds on a scheduled trigger. Deduplicates against previously seen items using a Postgres seen-items table. New items pass to the scoring stage; previously seen items are discarded.

Nodesn8n Schedule Trigger → RSS Feed nodes → deduplication check → new items queue
Stage 02

Write-Score Calculation

Each item is scored on four dimensions: (1) trend momentum — is this topic currently rising? (2) audience fit — does this match the target reader profile? (3) content utility — does this have practical value or is it noise? (4) monetization fit — does this create a natural path to a product or service? A risk penalty is subtracted for items flagged as sensitive, controversial, or legally ambiguous.

NodesLLM scoring node (structured output) → write-score calculation → score threshold filter
Stage 03

Routing — Alert or Archive

Items above the score threshold are routed to the Telegram trending alert. Items below threshold are written to the archive table for future reference but not surfaced. The threshold is configurable — raising it reduces volume; lowering it increases coverage.

NodesIF score > threshold → Telegram alert bot · ELSE → archive INSERT
Stage 04

WordPress Publishing (Optional)

High-scoring items can be optionally pushed to WordPress via REST API as draft posts — pre-populated with title, summary, source link, and score breakdown. Content team reviews drafts and publishes. Removes the blank-page problem from content creation.

NodesWordPress REST API → draft post creation → content team review queue
SECTION 04 / RESULTS
04 — Results

Active Build. Core Pipeline Operational.
Expanding to Client Deployments.

TheFeedLab.io is an active build — the core RSS aggregation and write-score pipeline is operational on the DEV n8n instance at dev.thefeedlab.io (Cloudflare tunnel). The Telegram trending alert bot is live and delivering scored signals.

The write-score algorithm has been calibrated through iteration on real feed content. The four-dimension scoring produces meaningfully different output from a simple recency sort — items that score high tend to have multiple monetization paths and low risk exposure, which is the profile that produces the best content ROI.

The pattern is directly applicable to client deployments outside content: review monitoring for retail operators, load board signal scoring for freight brokers, and brand mention intelligence for any business with an online presence. TheFeedLab is the reference implementation for the signal intelligence architecture.

Live
Core Pipeline Status
4
Score Dimensions
1
Cloudflare Tunnel Active
Feeds Configurable
Client patternSame pipeline deployed as review intelligence for retail · brand monitoring for food service · load signal scoring for freight
SECTION 05 / LEARNINGS
05 — What We Learned

Signal Quality Beats
Signal Volume, Every Time.

The risk penalty is as important as the score.

Items that score well on all four dimensions but carry legal, reputational, or political risk should not be surfaced. Subtracting a risk penalty rather than adding a risk dimension preserves the positive scoring dynamic while filtering dangerous content.

Deduplication is infrastructure, not a feature.

Without a reliable deduplication layer, high-volume RSS feeds produce duplicate alerts within hours. The Postgres seen-items table is not glamorous engineering — it's the thing that makes the system usable at scale.

The pattern generalizes beyond content.

Ingest → score → surface is the same architecture for review monitoring, load board scanning, and brand intelligence. TheFeedLab is the content implementation of a general pattern that GRS deploys across verticals.

Configurable thresholds matter more than fixed ones.

Different use cases need different sensitivity levels. A content team might want 10 items a day; a crisis monitoring team might want 100. Exposing the threshold as a configurable parameter rather than hardcoding it makes the system genuinely reusable.

FINAL CTA
Next

Want Signal Intelligence
for Your Operation?

Review monitoring, brand alerts, load board scoring — the same pipeline deployed for your specific use case. 30-minute discovery call.