Part of: Why your month-end close keeps getting slower
The dimensional framework decision is the single most consequential architectural choice in a multi-entity ERP implementation. It is also the choice most teams make casually, in week three of a six-month build, without realizing it determines what the business can ever report on for the next five to ten years.
The default modern ERPs offer is four to six dimensions: location, department, class, project, customer, vendor — or some variant. The team picks four because the implementation partner suggests it. Three years later, finance discovers it cannot run a product-line profitability report because product line was never dimensionalized. The fix at that point is a chart-of-accounts restructure that takes six months and a significant budget.
Five questions determine whether four or six is right.
Question one — how many ways will the CEO ever want to slice the income statement? Not how many ways the controller currently slices it. The CEO. By location, department, product, channel, customer segment — list them all, then design dimensions that cover them.
Question two — what reporting is required for lenders or investors? Debt service coverage by property, segment profitability for a PE sponsor, cost-center accountability for an operating partner. These requirements often reveal dimensions the internal team would not have chosen.
Question three — how disciplined is the team at consistent coding? More dimensions require more discipline. A team that struggles to code consistently against four dimensions will not handle six well.
Question four — what integrations will feed the ERP? If payroll, HR, CRM, and eCommerce platforms will push data in, their dimension structures need to map to the ERP. Designing the ERP dimensions without modeling the integrations first produces mapping problems discovered late.
Question five — what is the growth scenario in five years? The business growing from five to twenty entities has different dimensional needs than the business staying stable.
The answer is worth three weeks of careful design. It is not worth three months of restructuring.
Filed under
