All Docs
FeaturesAgentOS Scope OutUpdated March 15, 2026

Top-N Auto-Deep-Analysis Selector

Top-N Auto-Deep-Analysis Selector

After a skim crawl, the platform automatically identifies the most promising products and flags them for deep analysis — without requiring manual triage.

Overview

Once a directory skim completes, the system scores every discovered product using two readily-available surface signals:

  • Review count — a proxy for market traction and demand.
  • Pricing tier — a proxy for revenue potential.

The top N products (default: 20) are automatically flagged. You review the shortlist, make any adjustments, then confirm to launch the full deep-scrape pipeline against those products.

Workflow

1. Paste directory URL
       ↓
2. Skim crawl runs (all listings scanned for surface signals)
       ↓
3. Products ranked by review count + pricing tier
       ↓
4. Top 20 auto-flagged for deep analysis
       ↓
5. Review flagged list → adjust if needed → confirm
       ↓
6. Deep-scrape pipeline launches on confirmed products
       ↓
7. Full scored dossiers generated and ranked on dashboard

Configuration

The number of auto-flagged products is configurable:

ParameterDefaultDescription
topN20How many top-ranked products to flag after a skim crawl

Adjust topN to cast a wider or narrower net before the deep-analysis stage.

Flagging Signals

SignalWhat it measures
Review countMarket traction — products with more reviews indicate higher demand and user engagement
Pricing tierRevenue potential — higher-priced products suggest willingness-to-pay in the segment

These signals are deliberately lightweight — they are extracted during the fast skim pass without triggering a full deep scrape.

Review Step

Before the deep-scrape pipeline is triggered, you are presented with the auto-flagged list. At this point you can:

  • Inspect each flagged product's surface-level data.
  • Remove products that don't fit your research focus.
  • Add products from the broader skim results that the auto-ranker may have missed.
  • Confirm the final list to start deep analysis.

This confirmation gate ensures that expensive deep-scrape jobs are only run on products you actually want to analyse.

Relationship to Composite Scoring

The auto-flagging step uses a simplified surface score (review count + pricing tier) as a fast filter. Once a product passes through deep analysis, it receives the full composite opportunity score covering replicability, market demand, revenue potential, and competitive gaps — which drives the final dashboard ranking.