Insurance
BI modernisation and claims forecasting for a top-10 SA insurer
Month-end reporting cut from nine days to one; claims forecasting live.
9 days → 1month-end reporting cycle
1 sourceof truth replacing 400+ departmental workbooks
±8%claims-volume forecast accuracy at 90 days
The context
The insurer's reporting ran on hundreds of Excel workbooks fed by manual extracts from policy, claims and finance systems. Month-end consumed nine working days; numbers differed by department; and the executive team had no forward view of claims exposure.
The challenge
Create one governed source of truth across policy, claims and finance — and use it to forecast claims volumes — under POPIA and internal audit scrutiny.
What we built
- A dimensional data warehouse consolidating the three core systems, with automated, monitored ETL replacing manual extracts.
- Power BI dashboards with row-level security so brokers, managers and executives each see exactly their scope.
- A machine-learning claims-forecasting model surfaced directly in the executive dashboard, with documented assumptions and monthly accuracy review.
- Data governance artefacts — lineage, ownership, lawful-basis register — that passed internal audit first time.
Stack
- SQL Server
- Azure Data Factory
- Power BI
- Python (ML)
- Row-level security
“For the first time, exco starts the month arguing about what to do — not about whose numbers are right.”
Detailed client references are available under NDA on request.
Have a similar challenge?
This engagement was delivered by our Data & Business Intelligence practice.