What Your Numbers Say About Your Processes

The instinct when a cost line moves the wrong way is to discuss the cost line. Budgets are revised, targets are reset, and the mechanism that produced the number goes unexamined for another quarter. Financial statements are useful here for a reason that has little to do with accounting. They are the only complete record most organizations keep of where their effort actually goes.

Your accountant answers a different question

Accountants tell you what the numbers are and whether they are correct, which is essential and not what this is. The unanswered question is usually what the numbers are telling you about how the business runs.

Financial statements are organized for reporting: a chart of accounts, by category, by period. That structure is right for its purpose and wrong for finding out why margin fell, because it groups by type of expense rather than by the activity that caused it.

Where margin actually is, versus where you think it is

Start here, because it is where the surprises live. Margin by line of business, by service, by customer type - and frequently by individual customer, which is uncomfortable and useful.

The common finding is that the work everyone is proud of is subsidising something nobody examines. A service that looks profitable at gross margin turns out to consume a disproportionate share of senior attention, rework and expedite costs, none of which appear against it anywhere.

Which means the question is not “what does this earn” but “what does this consume”, and consumption is a process question.

The costs that have no line

Some of the most significant operational costs never appear as a line item anywhere, which is precisely why they persist.

Rework. Doing something twice because it was wrong the first time. It shows up as labour, indistinguishable from work.

Manual reconciliation. Somebody comparing two systems that do not talk, usually in a spreadsheet, usually the same person every week, usually because of a decision made years ago.

Waiting. Files sitting in queues nobody owns. Invisible in the accounts, plainly visible on a process map as the gap between two lanes.

Approvals that protect against nothing. A sign-off introduced after an incident in 2019 that has been applied to every transaction since, adding a day each time.

Expediting. The premium freight, the overtime, the discount to keep a customer whose order slipped. Usually recorded, rarely traced back to the step that caused the slip.

Reporting that measures something other than what people believe

Two checks worth running on any set of management reports.

First, for each measure, ask what specifically it counts and when the definition was last confirmed. Measures drift as systems change, and a metric everyone trusts is occasionally counting something quite different from its label.

Second, for each measure, ask what decision anyone has made because of it in the last year. A surprising proportion will have no answer. Those reports cost real time to produce and can be retired, which is usually the fastest win available.

From a number to a step

1. Pick the number that bothers you

   Pick one, not a dashboard: the number that moved in the wrong direction, or the one that never moves at all.

2. Ask which activity produces it

   Not which department - which activity. “Cost of sales” is a category; “we re-key every order into two systems” is an activity.

3. Map that activity

   From the people who perform it. Current-state mapping, ninety minutes a head.

4. Find the step

   It is usually a handoff, a manual bridge between two systems, or an approval nobody can justify.

5. Cost the step, then decide

   Now you have a number attached to a specific change, which is the only basis on which a process improvement should be approved.

Why a short review is often the cheapest way to start

A review of this kind frequently identifies which process is worth mapping first - which makes the larger engagement narrower, cheaper and better targeted than commissioning a mapping exercise across a whole function and hoping the important thing turns up.

Start with the number. Follow it to the step. Then do the process work with something specific in mind.