When product and customer profitability drifts from the general ledger

The most common failure is a profitability model that drifts from the general ledger. Allocation drivers go stale and untagged costs land in catch-all accounts, so product and customer margins mislead. Tying model totals back to ledger balances exposes it. Tagging costs at posting, with named driver owners, fixes it.

How it starts

It rarely begins with a bad decision. It begins with a reasonable shortcut.

The profitability model is usually built outside the ledger, in a planning tool or a large workbook. Someone designs it carefully, picks allocation drivers, and ties it to the books at launch. Then the business moves on. A new product line appears. A warehouse closes. A shared service team takes on work for a second division. None of these changes reach the driver tables, because nobody owns them.

Meanwhile the ledger keeps changing underneath. New cost centres get created without the attributes that the model reads. Accountants under pressure at close post manual journal vouchers to a general overhead account because it is quicker than finding the right product or customer code. Accruals get booked at a summary level and reversed later in detail, so the period the model sees is not the period that actually happened.

Asset costs add another layer. Depreciation and impairment sit in the fixed asset records, often grouped by asset category. If those charges reach the model only as one lump, life cycle costing for a product that depends on specialised equipment looks far cheaper than it is.

Each gap is small on its own. Together they shift cost away from the products and customers that consume it and spread it evenly over everything else.

How it gets spotted

The clearest signal is a reconciliation that no one performs any more. When the total cost in the model is compared with the total in the ledger for the same period, the two disagree, and the difference has been carried as an unexplained line.

Other signs show up in the business before finance notices. A product is discontinued for poor margins, yet the overhead it supposedly carried stays put. Sales teams argue that a large customer looks unprofitable only because it absorbs a share of costs it never touches. New product evaluations keep beating their targets on paper while the division as a whole underperforms.

Activity-based measures give it away too. If cost per order or cost per shipment barely moves across periods where volume changed sharply, the drivers are probably frozen.

Another clue is the size of the unallocated pool. When the catch-all account keeps growing and the model simply spreads it pro rata, the analysis has stopped saying anything about cause.

Questions to ask the people who run it

The documented process and the real one often part ways here. These questions tend to surface the difference.

• When the model total and the ledger total disagree, who notices, and what do they do with the gap?
• Which driver tables were last updated by someone still in the team?
• At close, where do journals go when the right product or customer code is unclear?
• Are any cost centres or accounts missing the attributes the model depends on, and how would anyone find out?
• Do depreciation charges arrive in the model by asset, by category, or as a single figure?
• When a strategy for a new customer segment is tracked, are its costs pulled from the ledger or estimated in a separate file?
• Has anyone outside finance challenged a margin figure, and what happened next?

The answers to the third and fifth questions are usually where the largest leaks sit.

How it is fixed

Start by restoring the tie-out. Every reporting cycle, the model should reconcile to ledger balances before any margin figure is released. Any residual needs an owner and an explanation, not a plug.

Next, move the coding upstream. Account and cost centre setup should require the attributes the profitability analysis needs, so a new centre cannot be created without them. Manual journals posted to general overhead should need a reason code and a reviewer who understands the allocation logic.

Give each driver a named owner in operations, not only in finance. The person running the warehouse knows when pick volumes shift. The person running the service desk knows which division generates the tickets. A short review with them each cycle keeps drivers grounded.

Bring asset cost through at the level of detail the products actually use. For equipment dedicated to a product family, depreciation should follow that family directly.

Finally, shrink the unallocated pool deliberately. Each item left there is a cost the analysis cannot explain, and every one removed makes the customer and product mix decisions more defensible.

What makes it come back

The repair decays the same way the original model did. Reorganisations, new systems and staff turnover all reopen the gaps. Continuous cost improvement work depends on this discipline holding, so the tie-out and driver reviews belong in the close calendar, owned like any other control.

APQC's Process Classification Framework® (PCF) is an open standard developed by APQC, a nonprofit that promotes benchmarking and best practices worldwide. To download the full PCF or to view definitions and measures, please visit www.apqc.org/pcf.