Work
Case study · Analytics & Forecasting

Forecasts the team can stand behind.

A Nordic manufacturer ran on five reporting tools and no two numbers agreed. We consolidated the data into one governed model — and month-end went from two weeks to two days.

SectorIndustrial manufacturing
EngagementDiagnostic + retainer
Timeline8 weeks to first dashboards
Project film
01
At a glance
40
report assembly handed back a month — about a working week (estimate)
2
month-end close — down from two weeks
1
single source of truth across the company
5%
forecast error against actuals
02
Challenge
The problem

Five tools, no shared truth.

Sales lived in the CRM, production in the ERP, finance in spreadsheets — and every Monday meeting started with an argument about whose number was right.

Reports were assembled by hand, arrived late, and were stale on arrival. Forecasting was gut feel with a spreadsheet attached.

  • Manual exports stitched together every week
  • No agreed definitions — “revenue” meant three different things
  • Forecasts nobody was willing to plan against
03
Solution
What we built

One model. One set of numbers.

We started by agreeing definitions — what counts, when, and who owns it. Only then did we build pipelines from the ERP, CRM and spreadsheets into one governed model.

Dashboards now read in seconds, forecasts are back-tested against history before anyone plans with them, and quality checks catch broken numbers before the team does.

  • Automated pipelines replacing weekly manual exports
  • A documented semantic layer — every metric defined once
  • Demand and cash forecasts validated against history
04
Result
The outcome

Two days to close. One truth.

Month-end close fell from two weeks to two days, report assembly disappeared from the calendar, and planning meetings now start from the same number.

“The Monday meeting stopped being about whose number is right.”

Background

Why the numbers diverged.

The company had grown product line by product line, and every team bought the tool that fit its own job. Each tool was right locally — and wrong globally. Nobody owned the whole picture, so finance reconciled it by hand every month.

Approach

How the eight weeks ran.

Weeks 1–3 — Definitions & mapping. We sat with every team, mapped where each number was born, and wrote definitions everyone signed off on.

Weeks 4–6 — Pipelines & model. Automated pipelines replaced the exports; the semantic layer made every metric mean one thing.

Weeks 7–8 — Dashboards & adoption. Dashboards built around real decisions, plus training so reporting became routine.

What's next

From reporting to planning.

With the model trusted, forecast coverage is expanding into capacity planning and purchasing. Because definitions are documented, each new question is a view on the same model — not another tool.

Have a challenge like this?

Make the numbers something you can plan on.

Book a diagnostic