Most FP&A models work well at small scale.
One business unit. One leadership team. One set of assumptions.
The problems start when the same model is stretched across multiple business units, regions, or entities. What breaks is not the math. What breaks is structure, ownership, and decision logic.
We see this pattern repeatedly across Adaptive, Anaplan, OneStream, and Excel-to-EPM migrations.
1. Assumptions Stop Being Comparable
At single-BU scale, assumptions feel universal.
Growth rate, attrition, price, productivity, FX, seasonality — all appear consistent.
Once multiple business units are added, those assumptions quietly diverge.
Each BU operates with different:
- Revenue drivers
- Cost behavior
- Hiring velocity
- Local seasonality
- FX exposure
The model still aggregates cleanly, but comparability is gone. FP&A ends up reviewing totals without confidence that inputs mean the same thing across units.
The result is false precision. Numbers look aligned, but decisions are not.
2. Workforce Logic Becomes Inconsistent
Workforce planning is usually the first visible failure.
One BU plans headcount by position.
Another plans by role family.
Another plans by cost center with manual overrides.
The model allows all three, because flexibility is seen as a feature.
At scale, that flexibility becomes a control problem. Attrition, backfills, open roles, and vacancy timing behave differently by BU, but roll up into one workforce view. FP&A can no longer explain why cost variances exist without deep manual reconciliation.
The issue is not data quality.
The issue is mixed planning logic living in the same model.
3. FX and Currency Handling Starts Distorting Signals
Single-BU models often treat FX as an afterthought.
Multi-BU models cannot.
Different entities transact in different currencies, plan in different base currencies, and report in a consolidated currency. When FX logic is layered late or handled inconsistently, forecast accuracy degrades silently.
We often see:
- Rates applied at the wrong grain
- Inconsistent “as-of” dates
- Mixed use of average vs end-of-month rates
- Adjustments done outside the model
The consolidated forecast still balances.
The drivers behind it no longer explain reality.
4. Ownership of Numbers Becomes Unclear
At small scale, ownership is implicit.
At multi-BU scale, it must be explicit — and often is not.
Who owns:
- Hiring assumptions?
- Productivity ratios?
- Price changes?
- Local adjustments?
When ownership is unclear, FP&A becomes the default owner of everything. Business partners disengage. Reviews turn into variance explanations instead of forward decisions.
The model still runs. The operating rhythm breaks.
5. Time Granularity Stops Matching Decisions
Single-BU models often plan monthly or quarterly without friction.
Multi-BU environments expose mismatches:
- Some decisions are weekly
- Some are monthly
- Some are annual
The model forces one cadence. BUs work around it with shadow models. FP&A reconciles after the fact.
The plan becomes a reporting artifact instead of a decision tool.
6. Aggregation Hides Local Reality
Rollups are comforting. They are also dangerous.
At scale, aggregation smooths volatility that matters locally. A hiring delay in one BU offsets early hiring in another. A pricing miss in one region is masked by performance elsewhere.
Executives see stability. Operators feel friction.
This gap erodes trust in the model.
7. Change Management Becomes the Bottleneck
Every new BU introduces:
- New planning behaviors
- New terminology
- New tolerance for process
Without a clear planning contract, each rollout increases entropy. FP&A spends more time maintaining the model than improving it.
At this point, teams say the tool is the problem.
In reality, the operating model is.
What This Means for FP&A and EPM Design
Scaling FP&A models is not a technical exercise.
It is an operating design problem.
What holds at one business unit does not automatically hold at ten.
Successful multi-BU models:
- Standardize planning logic before standardizing numbers
- Separate local flexibility from global comparability
- Treat workforce and FX as first-class design elements
- Make ownership explicit at every planning layer
When these principles are ignored, models still function — but decisions degrade.
That is what actually breaks.
