KPIs · Series “Reporting that carries decisions”, part 10
Customer profitability: which customers to let go
Every management team knows its customer list sorted by revenue. Sorted by contribution margin it looks different – and only then does it carry the decision about which customers to keep, restructure or let go.
Philipp Kempter · 5 October 2026 · 10 min readIn the appendix of the monthly management report sits the list of the twenty largest customers, sorted by revenue. The first name has been the same for years. That customer gets the most experienced sales representative, special terms and first call on capacity. When the controller sorts the list by contribution margin for the first time, the customer drops to sixth place. After the cost of serving them – never measured until now – the margin is close to zero. The head of sales asks, quite reasonably, what is supposed to follow from that number. That question is where a customer profitability analysis starts, not where it ends.
The decision that customer profitability has to carry
Before we define a single metric, we ask the purpose question: What are these numbers supposed to say, and what are they for? For customer profitability the answer is a bundle of three decisions. First, adjust terms: discount scale, minimum order value, freight contribution, payment terms. Second, adjust service: who gets field visits, who is served by inside sales or the web shop, who gets free technical advice. Third, let the customer go: to a distributor, a competitor or a process without personal account management.
For each decision the decision inventory settles the same three things. Who decides: terms sit with the head of sales within a framework, hand-overs with the executive board. How often: terms once a year with the pricing round, service and hand-over quarterly. What room to move: a customer with a framework contract until 2028 cannot be let go, only restructured – for that account the number is an analysis metric, not a steering metric.
The problem is not that these decisions go unmade. They are made – on the basis of the revenue list. Ossola-Haring, Schlageter and Schöning deliberately put the fallacy "revenue is the most important metric" first among their eleven: revenue alone says almost nothing, meaning only emerges together with contribution margin, customer share and days sales outstanding (Ossola-Haring et al. 2019, fallacy 1). A typical calculation shows the cost: a customer with CHF 4.8 million revenue, 22 % discount, 340 orders a year, monthly visits and 6 % returns delivers less contribution margin after freight, commission and service than a customer with CHF 1.9 million and 40 orders. And the capacity the big account ties up is missing at exactly those smaller, quieter customers.
The questions behind the decision
Two questions follow, and the report has to answer both. First: which customers still contribute after the cost to serve? Not after material cost – sales knows that margin already – but after everything the customer triggers that never appears on an order line. Second: where does small-order chaos eat the margin? In procurement controlling, Kunesch answers "how much does chaos cost?" with the share of rush and small orders (Kunesch in Eschenbach et al. 2021, p. 30). On the sales side it is the same question: a customer who orders the annual volume in 340 orders instead of 40 causes eight times the handling, picking, freight and invoicing – for the same revenue.
König's guiding questions for customer contribution accounting connect here: where does the lower profitability come from – cost structure or terms? Did the acquisition pay off? Which cost drivers are at work? (König in Eschenbach et al. 2021, pp. 615–618). They decide which level of contribution margin we need.
Metrics: the multi-level customer contribution margin
The answer is a family of metrics, not a single value. The base is the multi-level customer contribution margin. CM I: net revenue less material and purchase cost – the product margin from the customer's mix. CM II: less the costs directly attributable to the order – freight, packaging, commission, cash discount, customisation. CM III: less the cost to serve per customer – field visits, technical advice, returns handling, dunning, special logistics requests. Four supporting measures come with it: the customer contribution ratio per level (margin as a percentage of net revenue, because an absolute margin always favours large customers), the net revenue ratio (net to gross revenue, the terms in one number; König in Eschenbach et al. 2021, p. 618), order size (net revenue per order as a proxy for small-order chaos) and cost to serve per customer as its own line.
Six metrics are not yet steering. So we apply Kobi's filter: strategic relevance, informative value, controllability and availability (Kobi in Eschenbach et al. 2021, p. 373). Strategic relevance: CM II and CM III per customer carry the three decisions directly, CM I only the product-range question – it stays in the appendix. Informative value: ratio and absolute margin say different things, margin and capacity tie-up – both stay. Controllability: sales can move net revenue ratio and order size, not material cost, so no CM I value gets a traffic light. Availability: CM I and CM II sit in the profitability analysis, the cost to serve does not – which brings us to the data.
Weber and colleagues give a rule for how deep a metric tree should go: the breakdown ends where the measure gives direct guidance for action in daily business (Weber et al. 2017, p. 94). For the head of sales that is net revenue ratio and order size per customer; for the board it is CM III per customer. The report goes no deeper; order-line level is analysis, not reporting.
Data reality: cost to serve is not a standard report
We think of data as one large table of characteristics and value fields. Characteristics: customer, customer group, sales channel, entity, period, order. Value fields: gross revenue, discount, net revenue, material, freight, commission, cost to serve. A profitability analysis in the style of SAP CO-PA is exactly that: every invoice line with its characteristics and value fields in one table. CM I and CM II come out of it in one query, provided the customer characteristic sits cleanly on every line – and freight and commission are booked per order, not as a monthly lump sum on a cost centre.
The cost to serve is in none of those lines. Field sales logs visits in the CRM, if at all; technical advice appears on no invoice; returns run through the warehouse; special requests live in e-mails. Kunesch describes the same finding for inventory: the number you would need is usually not a standard report of the system (Kunesch in Eschenbach et al. 2021, p. 28). So the cost to serve has to be collected before it can be reported. The measurement plan says how: visits from the CRM, advisory hours from time recording, returns from the warehouse system, each valued at a standard rate. Not full costing, but an approximation with four drivers – Kobi calls it forgoing perfection (Kobi in Eschenbach et al. 2021, pp. 375–376).
And the value must reconcile. The sum of all customer CM II has to equal the contribution margin in the income statement, otherwise the meeting discusses the difference instead of the customer. Reconcilability before beauty: the report hangs on the close, valid from the fifth working day, and the cost to serve is shown as its own line so everyone sees what is accounting and what is estimate.
| Field | Content |
|---|---|
| Decision | Terms, service level, hand-over per customer – head of sales (terms), board (hand-over) |
| Question | Which customers still contribute after order and service cost, where does small-order chaos eat the margin? |
| Metric | CM II per customer, TCHF and % of net revenue; order size as additional column |
| Definition | Net revenue − material at standard − freight − packaging − commission − cash discount; per invoice line, aggregated by customer |
| Source | Profitability analysis (CO-PA style), customer master (group, channel); cost to serve from Q2 via CRM, time recording, warehouse |
| Cadence | monthly from 5th working day; steering meeting quarterly |
| Owner | Head of sales; data owner controlling |
| Threshold | CM II ratio more than 5 percentage points below customer-group median in two consecutive quarters; negative CM II: immediately |
| Response | Sales presents a proposal per customer: terms, service level or hand-over – with capacity effect |
The report: sorted by margin, not by revenue
The report is one page per customer group, in the spirit of IBCS®. Customers are sorted by CM II, not by revenue – that alone changes the discussion. Each row shows prior year, actual and budget, time runs left to right, and the variance sits as its own bar beside them, because the value matters less than its movement. The threshold is ±5 %, five percentage points for the ratio. Only a metric with a threshold gets a traffic light: CM II per customer yes, CM I no, cost to serve once a prior year exists.
One of König's warnings belongs in the footnote: period comparisons are distorted by changes in customer and channel structure (König in Eschenbach et al. 2021, p. 618). If a large account moves from direct business to a distributor, the group's margin falls without anyone doing a worse job. That is why the number of active customers per group sits next to the prior-year comparison.
Effect: threshold, owner, action – quarterly
A report does not steer; a response path does. Threshold: a customer's CM II ratio more than five percentage points below its group median for two consecutive quarters – or a negative CM II, then immediately. Owner: the head of sales for terms and service, the board for hand-overs. Action, named in advance by type: adjust terms, change service level or prepare a hand-over – each with the expected capacity effect in hours, because that is the counter-calculation: what happens to customers B to F when customer A ties up less attention?
The cadence is quarterly. Monthly, an individual customer's margin swings too much – one project order or one return shifts the ratio by percentage points, and Ossola-Haring and colleagues warn that control intervals that are too short turn noise into signal and trigger overreaction (Ossola-Haring et al. 2019). The report runs monthly with the close; decisions are taken quarterly. After four quarters comes the review: did the metric carry the decisions? Were customers restructured, and did their margin rise? Is the cost to serve reliable enough to make CM III a steering metric? And which customers have sat below the threshold for a year without anything happening – that is the line that says most.
At an industrial company in building technology, running SAP alongside a separate finance system, a CO-PA-style profitability analysis was in place and CM I and CM II per customer were cleanly available – nobody had ever sorted them by margin, only by revenue. The cost to serve existed nowhere: visits sat in the CRM, advisory hours in time recording, returns in the warehouse, none of it ever attributed to a customer. With sales I defined four drivers, set a standard rate for each and wrote the collection into the measurement plan – two quarters before CM III entered the report. Until then we worked with CM II and order size, and even that moved three of the ten largest customers by revenue into the bottom half. Reconciling the customer total with the income statement was the first condition the board set, and it was the right one: without it, sales would have questioned every number.
Gladen places the starting point for the objective information need not with the recipients but with the problems to be solved, derived from the company's goals (Gladen 2014, p. 7). For customer profitability that means: not "which customer metrics does sales want?" but "which problem are we solving – tied-up capacity, terms given away, small-order chaos?". Weber and colleagues add: metrics that hang in no driver hierarchy are to be questioned and otherwise removed from the reporting (Weber et al. 2017, p. 92). The top-20 revenue list hangs on none of the three decisions. It moves to the appendix.
What you can do tomorrow – and what the Scoping Sprint delivers
Have controlling sort your top-20 list by CM II – the number sits in almost every profitability analysis – and count how many customers move more than five places. Then ask sales how often it visits the largest customer and what that costs. The answer is usually the start of the measurement plan.
If you want to set the decision up properly, that is the work of the Scoping Sprint: in two to three weeks at a fixed price you get the decision inventory for terms, service and hand-over, the metric profiles for CM II and CM III, the measurement plan including the cost-to-serve collection, and the notation concept for the report. If you already have customer contribution accounting and want to know whether it carries the decision, the Project Review delivers a written second opinion with a cut list within a few days.
Sources
- Eschenbach, R., Baumüller, J., Siller, H. (eds.) (2021): Funktions-Controlling. Praxishandbuch für Unternehmen, Non-Profit-Organisationen und die öffentliche Verwaltung. 2nd ed., Springer Gabler – including Kunesch: Beschaffungs-Controlling, pp. 11–36; Kobi: Personal-Controlling, pp. 365–388; König: Vertriebs-Controlling, pp. 607–628.
- Ossola-Haring, C., Schlageter, A., Schöning, S. (2019): 11 Irrtümer über Kennzahlen. Mit den richtigen Erkenntnissen führen. 2nd ed., Springer Gabler.
- Weber, J., Bramsemann, U., Heineke, C., Hirsch, B. (2017): Wertorientierte Unternehmenssteuerung. Konzepte – Implementierung – Praxis-Statements. 2nd ed., Springer Gabler.
- Gladen, W. (2014): Performance Measurement. Controlling mit Kennzahlen. 6th ed., Springer Gabler.
- IBCS® Standards 1.2, © IBCS Association, www.ibcs.com, CC BY-SA 4.0.