Automated property deal-finder for a UK investor
A UK property investor buying, refurbishing and reselling homes around Solihull
A weekly pipeline that values around 33,000 property listings against official sales data and emails the best leads every morning.
The problem
The investor was searching property portals by hand for homes priced below market value. There were far too many listings to check properly, and good deals went to faster buyers.
The system now sweeps every listing inside a 60-minute drive each week, values each home against real sales nearby, and suggests an offer that still leaves the investor's target profit.
- 33,000
- listings valued every week
- 421 MB to 51 MB
- database size, to stay on the free tier
- 220+
- automated tests in Python and JavaScript
What I built
- Eight production n8n workflows covering ingest, clean-up, enrichment, scoring, the daily digest, a monthly open-data sync and alerts
- Custom scrapers on Apify, with every request going through a proxy and costs tracked per run
- Valuations from HM Land Registry sales and EPC floor areas, matched by property type and size
- A 0 to 100 score for discount, seller motivation, condition and risk, plus a suggested offer after stamp duty, finance, refurbishment and fees
- An Airtable lead CRM, a 07:00 email of the top 10 leads, and Telegram alerts if a run fails or costs climb
- GDPR controls designed in before any outreach: a legitimate-interest assessment, a privacy notice and do-not-contact suppression
Built against a written specification and an 18-milestone plan, with a plain-English system guide and an architecture diagram.
Tools: n8n, Apify, Crawlee, Playwright, Python, JavaScript, Supabase (Postgres), SQLite, Airtable, Telegram, Claude Code, MCP
Diagrams and screenshots
