Case study · Housing

Three merged housing associations, one lakehouse, £4.2m unlocked.

A G15-member housing group, newly formed from a three-way merger, needed a single data operating model. In 14 months we consolidated 48 source systems onto a Databricks lakehouse, retired three overlapping BI estates, and unlocked £4.2m of three-year savings.

£4.2m
3-yr NPV
48
Sources unified
11h → 3.5h
Batch window
−41%
Annual licence cost
The context

Three providers, three stacks, one regulator.

Following a three-way merger, the group had three QL / Capita / MRI estates, three BI teams (Power BI, Tableau, Qlik), and three operating models. TSM reporting required manual consolidation; damp-and-mould analytics were impossible at group level.

Databuzz was brought in — after two previous consultancies had stalled — to land a single operating model and an executable platform.

Approach

Strategy first, then migrate domain-by-domain.

01

8-week strategy

Target operating model, vendor selection, costed 18-month roadmap, board sign-off.

02

Foundation (months 1–3)

Databricks lakehouse, governance, identity, CI/CD, data-product templates.

03

Domain migration (months 3–12)

48 sources onboarded across tenancy, assets, repairs, finance, HR. Parallel-run, analyst enablement.

04

BI rationalisation (months 6–14)

Three BI platforms → Power BI on semantic layer. 72 curated dashboards retired to 28.

05

Governance & adoption (months 12–14)

Stewardship, TSM automation, regulator-ready evidence pack.

Outcomes, 14 months in.

£4.2m
3-year NPV

Licence consolidation, batch-window savings, analyst productivity.

−41%
Annual licence cost

BI and DW consolidation; Databricks replacing legacy MPP.

11h → 3.5h
Batch window

Re-engineered pipelines; incremental processing on Delta Lake.

0
Regulator findings

RSH visit cleared; TSM evidence automated.

96%
Board approval score

Quarterly programme reviews.

2.4×
Analyst productivity

Measured on time-to-insight baseline.

Technology

Stack.

Databricks
Azure
dbt
Power BI
Microsoft Purview
Fivetran
Great Expectations
Azure DevOps
FAQ

Common questions.

How was the programme governed?+
Weekly working group, fortnightly steering, monthly board report. Each domain had a business owner and a technical lead.
How did you minimise disruption?+
Strangler pattern by domain. The old systems kept reporting until each new domain was proven in parallel-run for 60 days.
What happened to the existing teams?+
Retained and reskilled. Our target from month 1 was capability transfer, not lock-in.
Ready when you are

Put your data to work.

Book a free 30-minute consultation with a senior Databuzz consultant.