Manufacturing · Case study

“Every report started with a manual export from SAP.”

A manufacturing and service group moved 40 reports off manual exports and onto one Microsoft Fabric model.

Based on actual client work. Client names and sensitive details are removed for confidentiality purposes.

40 → 0

reports fed by manual SAP exports

3

business units on one model

5

people on the client's data team run it

Based on actual client work. Client names and sensitive details are removed for confidentiality purposes.

The problem

The reports stopped when one person was away.

Every morning, one person exported SAP data into Excel files, and the reports read from those files:

When that person was away, the reports stopped

Nobody could say which export a number came from

So finance checked the important numbers again by hand before each meeting.

At a glance

Company

Company

Manufacturing and service group

Worked with

Worked with

Head of data, finance, the client's data team

Built on

Built on

Microsoft Fabric, Power BI, SAP as the source

What we did

What we did

Pipelines, lakehouse, one semantic model, report migration, training

A daily SAP export to Excel, run by hand by one person, ran on Monday and not from Tuesday to Friday that week. The Power BI production reports, from revenue and margin by plant to inventory days, all showed data four days old. 40 reports were not updated from Tuesday to Friday.
A daily SAP export to Excel, run by hand by one person, ran on Monday and not from Tuesday to Friday that week. The Power BI production reports, from revenue and margin by plant to inventory days, all showed data four days old. 40 reports were not updated from Tuesday to Friday.

What we built

Pipelines, one model, and a team to run it.

1

A foundation that loads itself

Nothing waits for an export anymore. One semantic model on top serves finance, sales and service.

Every night: pipelines load SAP into Fabric, the model refreshes early in the morning, and 40 reports are current by morning. No step needs a person.
Every night: pipelines load SAP into Fabric, the model refreshes early in the morning, and 40 reports are current by morning. No step needs a person.

2

Numbers checked against SAP

Before an old report was switched off, its totals were compared with SAP. A report moved only when every total matched.

Revenue by plant for March, checked against SAP: SAP shows 12.48 million euros, the old Excel report 12.31 million, 170 thousand short because one export was missed, and the new model 12.48 million, which matches. All 40 reports were checked this way before switch-off.
Revenue by plant for March, checked against SAP: SAP shows 12.48 million euros, the old Excel report 12.31 million, 170 thousand short because one export was missed, and the new model 12.48 million, which matches. All 40 reports were checked this way before switch-off.

3

A team that runs it

The client's five-person data team runs the pipelines and the model. We trained them on the new setup at handover, and every change goes through test before it reaches a report.

The client's five-person data team and what each person owns: two data engineers on the pipelines, an analytics engineer on the semantic model, a BI developer on the reports and a team lead on releases.
The client's five-person data team and what each person owns: two data engineers on the pipelines, an analytics engineer on the semantic model, a BI developer on the reports and a team lead on releases.

The outcome

Forty reports, one model, no manual step.

The reports no longer depend on one person.

The reports no longer depend on one person.

Every number can be traced back to SAP.

Every number can be traced back to SAP.

Copilot and the finance team read the same numbers from the model.

Copilot and the finance team read the same numbers from the model.

Show us the report nobody trusts.

1 · A 30-minute call.

2 · We look at your current reports together.

3 · We tell you what we would do.

Show us the report nobody trusts.

1 · A 30-minute call.

2 · We look at your current reports together.

3 · We tell you what we would do.

Show us the report nobody trusts.

1 · A 30-minute call.

2 · We look at your current reports together.

3 · We tell you what we would do.

CaseWhen is a Berlin BI consultancy that builds reporting that leaders can trust, on the Microsoft stack: Power BI, Fabric and Azure.

Berlin, Germany

© CaseWhen Consulting GmbH

CaseWhen is a Berlin BI consultancy that builds reporting that leaders can trust, on the Microsoft stack: Power BI, Fabric and Azure.

Berlin, Germany

© CaseWhen Consulting GmbH

CaseWhen is a Berlin BI consultancy that builds reporting that leaders can trust, on the Microsoft stack: Power BI, Fabric and Azure.

Berlin, Germany

© CaseWhen Consulting GmbH