Azure & data
Microsoft Azure
The cloud platform underneath: compute, managed data services, networking and the governance model around them.
Financial Services · Data, AI & Integration
Fifteen data sources into one lakehouse, and models that finish
A large member-owned credit union, United States. A substantial data footprint was spread across more than fifteen disparate sources, which made personalised member experience impossible and slowed every business decision that depended on analysis.
The situation
What the organization was dealing with
A substantial data footprint was spread across more than fifteen disparate sources, which made personalised member experience impossible and slowed every business decision that depended on analysis.
Member-owned institutions compete on relationship rather than on price, which means personalisation is not a marketing nicety but the actual product. Doing it requires knowing what a member holds across every system, and most institutions of this size have accumulated more than a dozen.
What was done
The work, and the part that was actually hard
Working with a Microsoft partner, data from all sources was centralised into a lakehouse on Microsoft Azure, becoming a single analytics platform for the institution.
Centralising fifteen-plus sources into one lakehouse is the unglamorous majority of this work. The headline — an 89% reduction in model run time — is a consequence of that consolidation rather than of a modelling improvement: analysis gets faster when it stops waiting for data to be assembled first.
Results
What was published
These are the figures exactly as reported in the source. Nothing has been rounded, extrapolated or restated.
Reduction in model run time89%
Sources centralised15+
Analytics platform1
What changed
- More than fifteen disparate sources unified into one lakehouse
- Model run time cut by 89%, changing what analysis was practical
- Personalised member experience built on a complete data picture
- Business decisions informed by analytics rather than by extract reconciliation
Platforms involved
What each product was doing here
Azure & data
Microsoft Fabric
One analytics platform over OneLake, so every workload reads the same copy of the data instead of its own extract.
What transfers
If you were to attempt this
The measurement is worth copying: run time on a real model is a better proxy for analytical capability than any platform capability list.
Where it usually gets harder than expected: A lakehouse that nobody has agreed the definitions for produces disagreement faster than the old extracts did. The governance conversation about what a member, a household and a product actually are is the part that determines whether this holds.
How we would take it on
Our approach to Data, AI & Integration work
1
Agree the definitions first
The least technical and most important step. If two functions cannot agree what a customer is, no architecture will reconcile their reports.
2
Deliver one workload end to end
Ingestion through to a report somebody actually uses, before the second workload starts. Platform builds that produce nothing for nine months lose their sponsor by month six.
3
Prove lineage
Every published figure traceable back to its source transaction. In a regulated setting that is a requirement; everywhere else it is what makes the number trusted.
4
Watch the consumption cost
Capacity-based compute behaves differently from licensed software. Monitoring goes in with the first workload, not after the first invoice.
Recognise the problem?
If any of the above describes your organization, tell us where it hurts most. We will tell you what the same platforms could realistically do in your environment, what we would measure, and whether we think it is worth doing at all.
Talk to our teamMore Data, AI & Integration storiesWe reply to every message within one business day.
Related reports
Others you may want to read
One view of the member, on every channelAlso in Financial ServicesThousands of service interactions a day, forty per cent automatedAlso in Financial ServicesScaling scarce expertise instead of hiring itAlso in Financial ServicesForty thousand web pages, answered in secondsAlso in Data, AI & Integration