British Housing: From CRM to Business Management System
The system now performs roughly 3,000 hours of work every month, the equivalent of about eighteen full-time staff. What began as a CRM became the system the business is run from.
British Housing is a national housing provider working with thousands of landlords and local authorities across the United Kingdom. An operation of that spread generates a great deal of information about itself and, in the ordinary course of things, sees very little of it.
The position was the one most businesses of this size will recognise. Enquiries arrived through several channels and were worked by hand, property and market information sat in systems that had no knowledge of one another, and the judgement that determines whether an opportunity is worth pursuing lived in the operator’s head. Any question worth asking took the better part of a day to answer, with the result that most were never asked at all.
The obvious remedy was a CRM. The remedy that was actually required turned out to be considerably larger, and the distance between those two things is the substance of this engagement.
The system now performs roughly 3,000 hours of work every month, which is the equivalent of about eighteen full-time staff across marketing, sales, data science and finance.
The application
A purpose-built CRM came first, covering enquiries, properties, landlords and local authorities together with the sequences that progress them and the states that track them. It was built to purpose instead of configured from a template, because the qualification logic here is particular to the business and no off-the-shelf pipeline models it.
It was delivered in weeks, is owned outright, and carries no per-seat licence. The web application and mobile experience sit on a bespoke design system with its own token layer and build pipeline, which is what kept the interface coherent as scope expanded well past the original brief.
The data layer
Beneath the application sits the part that changed the business. Automated pipelines ingest, normalise and resolve property and market information from live daily proprietary sources, built by hand for this client and available to nobody else.
The London dataset alone holds more than 700,000 property records. Each carries between twenty and sixty attributes before enrichment, and each is then resolved against energy performance, licensing, ownership and transaction records contributing substantially more, with the whole refreshed daily. The resolved dataset runs to over 640 million individual data points. That figure describes London, and the system is not limited to it.
Alongside it sits coverage of more than 130 local authorities and government contracts, maintained in real time.
The intelligence layer
An ontology and knowledge graph sit above the data, linking properties, landlords, local authorities, geographies and market comparables so that the system answers relational questions instead of returning rows. A question that would previously have occupied a week of manual work resolves as a single query, which changes not only how long an answer takes but how many questions get asked in the first place.
The assistant
An internally aware assistant runs across the whole system, holding context on the business and able to act within it. It answers questions against live enquiry, property and market data, manages the diary and books against it, surfaces suggestions on enquiries and properties, and carries out actions the operator would otherwise perform by hand.
It is wired into the ontology and the data path, which is what separates it from an assistant bolted onto a document store. It knows what the business knows.
What it became
Scope moved past sales some time ago. The system now carries finance, market intelligence, sales and the data science feeding all three, having stopped being a CRM and become the business management system the company is run from.
What it produced
The system performs roughly 3,000 hours of work every month across marketing, sales, data science and finance, which is the equivalent of about eighteen full-time staff carried over a year. That is the figure that reframes the investment, because the work being done is not work an existing team was doing faster. It is work the business would otherwise have had to hire for, carried out against data no competitor holds.
Leads rose by more than 600 per cent week on week in the first week of August 2026, and conversions rose by 30 per cent. Revenue is up 7 per cent month on month, following an increase on the month before that.
Leads up more than 600 per cent week on week. Conversions up 30 per cent. Revenue up 7 per cent month on month, on a base that had already risen.
Why it worked
The client could have bought a CRM from any number of vendors. What they would not have arrived at is the ontology underneath it, or the question of what the joined data could answer once it existed, because that question only occurs to somebody who has built the thing before and knows what becomes possible on the other side of it.
Engineers build what is specified. The value here lay in specifying the right thing, and then in the client owning the result outright.