02 · Dusted · Platform
Elevate
Dusted's proprietary market-research platform, analysing competitors across UX, ESG, martech, social, SEO and security. Automates 60+ digital datapoints and cut research time by ~80%.
- Role
- Full-stack engineer
- Client
- Dusted
- Timeframe
- 2023 — present
~80%
competitor research time cut
60+
datapoints automated per company
7
analysis domains per Elevate score
Overview
The story
Elevate is Dusted's proprietary market-research platform: point it at a company and it scores them across seven analysis domains — UX, SEO, security, social, ESG, martech and content — rolled up into a single Elevate score. Analysts group companies into industry collections of 100+ and compare them side by side.
Behind each score sits an automated data-gathering engine I built in Node.js: a headless-browser crawler that screenshots and analyses each site, enriched by third-party APIs — PageSpeed, accessibility auditing, SEO and backlink data, security scanning, social platforms, Trustpilot and company enrichment — with GPT-4o vision reading the screenshots for brand, messaging and ESG signals. Over 60 datapoints per company, gathered without a human in the loop.
The pipeline runs as a multithreaded worker pool on a VPS, fed by a job queue built natively in PostgreSQL — atomic task leases, heartbeats and watchdogs that respawn hung workers — so it survives crashes without any extra queue infrastructure. Results land in Postgres and surface through a React/TypeScript front end with score dashboards, per-domain analytics tabs, comparisons and shareable reports.
What used to be weeks of manual competitor research became an on-demand report. Elevate is now a key tool for client acquisition and pitch preparation across the agency — and it keeps growing, most recently with an AI feature that synthesizes best-practice page wireframes from an entire sector's navigation patterns.
Highlights
What makes it tick
- One composite Elevate score, decomposed into seven analysis domains per company
- 60+ datapoints gathered automatically — from Lighthouse scores to carbon footprint
- Headless-browser crawling layered with 8+ data APIs and GPT-4o vision insights
- Postgres-native job queue with atomic leases, watchdogs and crash recovery — no Redis required
- Industry collections of 100+ companies with side-by-side comparison and shareable reports
- AI wireframe generator that distils best-practice layouts from a whole sector
Stack
Built with
- React
- TypeScript
- Node.js
- PostgreSQL
- Puppeteer
- OpenAI GPT-4o
- Supabase
- AWS S3
Gallery
In the wild



