Strategy Sprint
Use-case prioritization, data-readiness assessment, roadmap and business case. Management decision: SPoT before AI pilot.
Case · Food production · 600 employees
8 reporting sources became 1 Single Point of Truth in 14 weeks
How a mid-sized food manufacturer restructured its data landscape so that reporting, steering and AI now work on the same foundation.
−68%Reporting effort
8 → 1Data sources consolidated
14 wksTime to production
< 12 mo.Payback period
The starting point
Eight sources.
Eight sources.
A nine-day month-end close
A growing company, a system landscape that had grown along with it, years of spreadsheet workarounds in controlling and business units. The AI initiative was on the agenda, but nobody knew what foundation it should build on.
Sales, production and finance metrics were spread across eight different sources: ERP, MES, three spreadsheet-based reports, a BI tool, manual consolidation in controlling sheets, and an additional tool from sales.
The effect: month-end closings took nine business days. Management received KPIs with a two-week lag. In controlling, manual consolidation tied up a third of the team's capacity.
The planned AI initiative, specifically a forecasting model for production planning, was up in the air. What data should the model even build on, when the sources contradicted each other?
Approach
Three phases.
Three phases.
14 weeks total
We start with a Strategy Sprint to clarify the levers, before anything gets built. Only then does the consolidation begin, followed by the rollout across business units.
Data consolidation
SPoT architecture, modeling with dbt, data-quality tests, migration of existing reports onto the new foundation.
Rollout & enablement
Training for controlling and business units, handover of the architecture to internal IT, launch of the AI pilot on a reliable foundation.
Result
What it concretely
What it concretely
changed
Operational and strategic. Both measurable in the first quarter after rollout.
Operational
- Month-end close shortened from 9 to 3 business days
- KPIs available in real time (previously: a 2-week lag)
- 1 source instead of 8, one consistent data model
- Controlling capacity freed up for analysis instead of consolidation
Strategic
- The AI pilot could start, on a reliable data foundation
- Forecasting model with measurable results in the pilot business unit
- Investment fully paid back in under 12 months
- A scaling foundation for further business units is ready
What we take away
Three insights
Three insights
that hold up again and again
Lessons learned from this project, ones that, for us, weren't a one-off, but recur in many similar situations.
Data consolidation before an AI pilot pays off
Trying to train AI on contradictory data usually costs more time than structuring the sources upfront.
Acceptance hinges on visibility
As soon as business units see "their" KPIs in real time, resistance to the architecture change drops significantly.
A clear format beats a timesheet
A clearly scoped project forces prioritization and delivers measurable results instead of open-ended consulting loops.
Keep reading
Related cases
Related cases
& insights
If this case resonated with you: these go deeper.
Facing a
similar situation?
In a 30-minute discovery call we jointly check whether a Strategy Sprint fits your situation. No pitch, no sales theatrics.