Group Forecasting Without Clean Data: Why Your Consolidated Result Is Always a Surprise

The question is not whether your group has a forecast. It does. The question is why the consolidated result is still a surprise, quarter after quarter, in both directions. The answer is almost always upstream of the model. It lives in the data that feeds the model, the timing at which that data arrives, and the intercompany flows that were never eliminated from entity-level submissions before they were aggregated into the group view. The forecast model is not the problem. The infrastructure feeding it is.

Why the consolidated result is systematically surprising

A Deloitte study of forecast accuracy in multi-entity European groups found that groups with five or more entities experienced forecast errors at the consolidated level that were, on average, 2.4 times larger than the errors in any individual entity forecast. The source of the amplification: inconsistencies between entity forecasts that become visible only when they are aggregated. Entities that use different pipeline definitions, different revenue recognition assumptions, and different cost allocation bases produce forecasts that are each internally coherent but that combine into a consolidated view that inherits and compounds all the inconsistencies simultaneously.

This is not a forecasting skill problem. Finance teams at the entity level are typically doing disciplined, careful work. The problem is structural: the inputs to the group forecast have not been standardised to a level that allows meaningful consolidation. The aggregation step transforms a set of individually reasonable entity forecasts into a group forecast whose reliability is materially lower than any of its components.

Four structural forecasting failure modes in multi-entity groups

Failure mode 1: Entity forecasts that are not comparable. When entities use different revenue recognition policies, different cost allocation methods, and different definitions of committed pipeline versus probable pipeline, the group forecast is an aggregation of incompatible inputs. A Group CFO who adds four entity forecasts and calls the sum the group forecast has produced a number that is internally consistent at no level. The entity forecasts are each internally consistent by their own definitions. The group forecast inherits the inconsistencies and presents them as a single figure without disclosure.

A professional services group, EUR 110M revenue, 6 entities across four markets, discovered during a finance transformation engagement that three entities included probable pipeline - deals at 50% or above in confidence - in their revenue forecast while two entities included only committed backlog. The difference in methodology produced a systematic optimism bias in the entities using probable pipeline. The consolidated forecast consistently exceeded actuals by EUR 2-4M per quarter. No entity's forecast was wrong on its own terms. The group forecast was structurally biased because the terms were not the same across entities.

Failure mode 2: Intercompany flows not eliminated in the forecast. Intercompany revenue and cost forecasts that are not eliminated at group level make the group result look larger than it is. In a group with significant intragroup activity - shared services charging growing as the group professionalises, management fees increasing as new entities are added, intragroup platform costs scaling with revenue - the unadjusted consolidated forecast shows revenue and cost growth that does not represent external commercial activity. Group revenue appears to be growing. The actual growth in external revenue is smaller. The difference is intragroup flow that should have been netted before the group view was assembled.

Failure mode 3: Timing misalignment across entity forecast submissions. Entities submit forecast updates at different points in the month or quarter. A group forecast assembled from submissions that are one to three weeks apart is not a group view. It is a patchwork of different moments in time. An entity whose forecast was submitted two weeks ago and that has since lost a material contract, experienced a customer payment delay, or identified a cost overrun is represented in the group forecast by numbers that no longer reflect its actual forward position. The Group CFO presents a forecast to the board that was accurate at the time of its components' submission and inaccurate as a group view at the time of presentation. The miss at quarter-end is structurally guaranteed.

Failure mode 4: Fixed cost misallocation distorting entity-level forecasts. Group overhead and shared service costs allocated to entities using static or arbitrary methods distort entity-level forecasts in ways that compound into the group forecast. When actual costs come in differently from forecast - because the allocation assumption was wrong, because group overhead grew faster than plan, or because the allocation basis changed mid-year without updating forecast assumptions - the group forecast variance is unexplained at entity level. No entity owns the variance because no entity owns the allocation methodology that produced it. The variance sits in the group consolidation, unexplained, until the audit flags it.

What a trustworthy group forecast infrastructure requires

The structural work that makes group forecasting reliable is not a forecasting methodology question. A group can use scenario-based planning, driver-based forecasting, zero-based budgeting - the methodology choice is less important than the infrastructure that executes it. Four infrastructure requirements are non-negotiable before a group forecast can be genuinely trusted.

Common close dates and submission deadlines across all entities - so the group forecast represents the same moment in time for every entity in the consolidation. Not different submission windows for different entities, not a rolling submission process that produces a patchwork. One deadline, enforced for all entities, producing a group view that is a group view.

Common definitions for revenue, pipeline, cost categories, and margin - so entity-level forecasts are comparable before they are aggregated. Not definitions that are similar but not identical. Identical definitions, encoded in the planning templates and the planning system, enforced by Group CFO governance.

Automated intercompany elimination in the forecast model - not a manual adjustment applied after the entity submissions have been assembled. A system-driven process that identifies and eliminates intercompany flows before the consolidated view is produced. This is buildable with most modern FP&A platforms. What makes it rare in practice is not technology capability but the absence of the clean IC data that the automation requires.

A single group forecast model, not entity-level spreadsheets aggregated manually. The group view produced by a defined process rather than by a person who knows how to run the process. When the person changes, the model continues. When the model is audited, it can be traced. When assumptions change, they can be updated centrally and the group view regenerates automatically.

The profitability forecast and the cash flow forecast: both, on the same foundation

In a group with material intercompany loans, dividend flows between entities, and cash structurally trapped in specific markets, the relationship between reported group profit and actual group cash generation is not straightforward. A group forecast that produces only a P&L view is incomplete, regardless of its accuracy. The Group CFO needs a cash flow forecast that starts from the same data foundation and accounts for entity-level cash positions, intercompany loan repayment schedules, and the FX impact on cash held in non-functional currencies. These two forecasts - profit and cash - must be produced from the same model, on the same assumptions, to be mutually consistent and both trustworthy.

Groups that have built this infrastructure consistently report a material improvement in forecast accuracy at the consolidated level - not because they hired better FP&A professionals, but because the professionals they had were working with better inputs. The infrastructure investment is fixed. The accuracy improvement compounds over time as entity-level data quality improves, automation reduces manual error, and the group forecast becomes a document that the board treats as authoritative rather than directional.

If your consolidated result is consistently surprising - in either direction - despite a functioning forecast process at entity level, the problem is almost certainly upstream of the model. Talk to us about what a group forecasting infrastructure review involves and where the structural fixes are most likely to close the accuracy gap.