Business Insights — Marshall & McAdam Dynamics

Live Python · FastAPI · SQLite · Playwright · Docker

Python FastAPI Playwright SQLite NLP Market Analysis Docker

How it works

01 — SCRAPE

Playwright crawls Google Play Store reviews by market category

02 — CLASSIFY

Reviews bucketed into Feature Gap, Outdated, UX, Bug, or Monetization

03 — SCORE

Multi-factor model weighs rating gap, staleness, install base, and review mass

04 — SURFACE

Ranked opportunity radar with filters, charts, and HN signal overlay

Live Demo

← Back to Projects

Build Journal

Summary

A market intelligence tool that turns Google Play Store complaints into a ranked shortlist of software opportunities. Playwright scrapes reviews by niche category, an NLP pipeline classifies pain points, and a multi-factor scoring model surfaces the highest-value gaps — zombie apps and broken feature promises score highest.

Implementation

Playwright scrapes app listings and reviews from Google Play by market category; results persisted to SQLite
Review text classified into six buckets (Feature Gap, Outdated, UX, Bug, Monetization, Unknown) using keyword + LLM heuristics
Opportunity score computed from rating gap vs category average, days since last update, install tier, review mass, and bucket weight
HN (Hacker News) signal layer overlaid — startup buzz in a category raises the score for incumbents with low ratings
FastAPI backend serves the scored dataset; Chart.js dashboard renders radar, bar charts, and a filterable opportunity table

Troubleshooting

Google Play serves bot-detection challenges — Playwright runs with realistic browser fingerprints and randomised scroll timing
Proxy path-stripping broke dashboard fetch calls — changed all API URLs from absolute to relative paths
Install count displayed as "1M+" strings — wrote a parser to convert display strings to numeric tiers for scoring
Scoring rewarded large apps by default — reweighted to favour the gap between user need and developer attention, not raw installs

To Do

[ ]Add App Store (iOS) as a second data source for cross-platform signal
[ ]Replace keyword classifier with a fine-tuned sentence transformer
[ ]Email digest — weekly top-5 opportunities delivered automatically
[ ]Competitor mapping — show which apps dominate each Feature Gap bucket
[ ]Export scored dataset to CSV for offline analysis