The spreadsheet that took three people to maintain
Limes House Properties managed multiple property locations with a fully manual workflow: staff monitored Rightmove, Zoopla, and niche portals daily, copied new listings into spreadsheets, formatted them for the website, and updated the CRM. Every hour spent copying data from portals was an hour not spent on clients.
We were losing time we didn't have. Every hour spent copying data from portals into our systems meant an hour less with clients.
What we built under the hood
The website looks straightforward — search, filter, enquire. Behind the scenes is a continuous data pipeline with four components:
- Multi-source ingestion. Scrapers and API integrations pull from multiple portals and normalize different formats into one unified schema.
- Deduplication and conflict resolution. Fuzzy matching identifies duplicate listings across platforms with confidence scoring — no property appears twice.
- Real-time search with ElasticSearch. Sub-200ms compound filtering for area, price, bedrooms, and property type.
- AI voice assistant. Handles initial enquiries and routes complex requests to human agents — available around the clock.
What changed for the business
Three staff members shifted from data entry to client-facing work. Property updates now reflect on the website within minutes of any status change. Fewer wasted viewings from stale or inaccurate listings.
Our website used to be a brochure we updated when we remembered. Now it's the most reliable source of our portfolio information.
The principle
Fixing a data pipeline doesn't sound exciting. But when you remove the daily grind of manual data entry, three people suddenly have time to do the work they were actually hired for — selling, advising, building relationships. Technology should operate invisibly.