What a single case really costs: the calculation behind it

Cost accounting in mid-size companies is built around products, orders and cost centres. It answers what a component costs or which department consumes which budget. A case, however, runs across all of that: an incoming invoice touches goods receipt, the specialist department, accounting and the management. Each of these stations books its time to its own cost centre, nobody books it to the case. The total therefore exists nowhere, even though every single element has long been recorded.

The second reason is the invisibility of manual work inside paid working hours. Typing a figure from a document into a second system generates no invoice that anyone has to approve. The effort only becomes noticeable when it turns into overtime or an additional post, and at that point it is usually explained as growth rather than as a process problem. That is exactly why costing pays off: it brings to light an effort that is already being paid for.

The third reason is respect for the effort involved. Many managers assume a process costing exercise means a project with consultant days, questionnaires and workshops. It does not. To decide which workflow to tackle first, a one-page calculation that combines three figures and two surcharges is enough. Everything beyond that is precision which does not produce a better decision at this stage.

The most common mistake in costing is an hourly rate that is too low. Anyone calculating with gross pay leaves out employer contributions, holidays, sickness, public holidays and everything that makes a workplace operational in the first place. According to the labour cost survey of the Statistisches Bundesamt, labour costs per hour worked in the German private sector averaged 45.00 euros in 2025 (Statistisches Bundesamt), with clear differences between sectors. That figure already includes non-wage labour costs, but no material or overhead costs.

For a process calculation the full-cost rate is the right measure: personnel costs plus a share of workplace, IT equipment, floor space, insurance and administration. In many companies this rate is noticeably higher than the pure labour cost figure. It is usually already available from cost centre accounting or can be obtained from the tax adviser. If no sound rate exists, take a conservative value and document it as an assumption rather than postponing the calculation because of it.

A second point concerns who actually handles the case. Approvals frequently land with managers whose rate is considerably higher. If the management reviews every incoming invoice individually, that minute belongs in the calculation at the corresponding rate. Points like this surface during costing before anyone talks about technology, and sometimes a simple value threshold for approvals is enough to relieve them.

Case volume is the easiest figure to obtain, because most systems already count. Document number ranges, order numbers, delivery notes, outgoing mail records, folder contents, form submissions on the website: an annual volume can be derived from any of these sources. Where no system counts, a tally sheet over the same period as the recording, extrapolated to the year, will do.

When extrapolating, seasonality is the main trap. Counting during a quiet summer month and multiplying by twelve underestimates the effort considerably. It is better to use the previous year's total, adjusted by a rough growth assumption. For the ranking the order of magnitude is enough: whether a workflow occurs two thousand or two thousand two hundred times changes nothing about the decision.

Once the case volumes have been gathered, it is worth not losing them again. A plain overview with volume, time and cost per workflow is the entry point into ongoing metrics and reporting, which later makes it possible to demonstrate whether a change actually worked. Without a baseline, every subsequent improvement remains an assertion.

Handling time and throughput time are two different measures. A case that consumes nine minutes of work can still spend eleven days in the building, because it waits for an approval, sits in the mailbox of an absent colleague or waits for the next batch run. The difference is idle time. It costs no direct wages, but it carries price tags: missed early payment discounts, late quotations, tied-up capital, additional queries about the current status.

Idle time is measured with two timestamps: receipt and completion. Both are already available in most systems, otherwise two date fields on the recording sheet will do. It is valued through concrete consequences: how many invoices were paid without the early payment discount, how many quotations went out later than promised, how often somebody had to ask about the status. These consequences can be quantified, the idle time itself cannot be directly.

Idle time is nevertheless one of the most effective levers, because it can often be shortened without any technical change: a value threshold above which a second approval is required, a deputising rule for absences, a fixed date instead of a batch run at the end of the month. Such measures cost nothing but a decision and take effect on throughput time immediately.

The most common objection is that the figures are not exact enough. That is true, but it is not a counter-argument, because the calculation has a different job from the annual accounts. It is meant to produce a ranking: which of the eight candidate workflows gets tackled first? That question is robust against imprecision. Even if every input figure is off by a third, the most expensive workflow will as a rule still be the most expensive one.

It therefore makes sense to work with ranges rather than point values: a cautious and a generous variant per workflow. If the cautious variant of one workflow is still above the generous variant of another, the order is settled. If the ranges overlap, the order between those two workflows does not matter anyway, and you take the one that is easier to implement.

There is a practical side effect too: people accept a rough, comprehensible calculation more readily than a model nobody can check any more. Every assumption fits on one page, and each of them can be challenged and corrected individually. When deciding on an investment, that traceability is worth more than decimal places whose origin nobody can explain three months later.

A handful of mistakes recur when costing workflows. They rarely overturn the ranking, but they cost credibility as soon as somebody checks the arithmetic. Knowing them avoids the usual objections in the discussion with the management or with the departments concerned.

One final note on the recording itself: if personal data is collected or conclusions about the performance of individual employees become possible, data protection requirements and possibly codetermination rights apply. In practice this can usually be avoided by recording without names and evaluating solely by work step. Assessing the specific case remains a matter for professional and legal review within the company.