Publicité
ERP IMPLEMENTATION
🇫🇷 Lire en français →

Cost-to-Serve in ERP: Measuring the Real Cost of Serving Each Customer and Channel

Cost-to-serve in ERP: 6 cost families, a 5-step method, ABC vs time-driven ABC, a worked example and the pricing and terms decisions it supports.

Cost-to-Serve in ERP: Measuring the Real Cost of Serving Each Customer and Channel

Two customers buy the same product range, at the same price list, for the same revenue. In your reporting, their gross margin is identical. Yet one places a single order a month, pays in 30 days and never calls. The other places twenty small orders, demands custom labels, regularly returns goods and settles in 90 days. To the finance team, the first is a good customer and the second needs a rethink. But only a cost-to-serve calculation can prove it, using your own company’s figures.

Cost-to-serve measures all the costs a company incurs to serve a customer, a channel or a segment, over and above the cost of goods sold. This guide is written for CFOs, financial controllers and managing directors of mid-market distribution and manufacturing companies. It explains how to build the model from data your ERP already holds, without a disproportionate tooling project, and above all how to turn it into commercial decisions.

It complements two other articles on customer margin: our guide to customer profitability analysis and gross margin in ERP covers the reporting side. Here, we focus on the method for allocating activity costs and on the decisions that follow.

Why Gross Margin Lies: The “Profitable” Customer Who Costs More Than They Earn

Gross margin answers a product question: how much is left after the purchase or production cost? It says nothing about what it took to get that product to that customer, or what happened after delivery.

In most mid-sized companies, these costs do appear in the books, but by nature (warehouse wages, freight, financial charges) or by cost centre (logistics, customer service, sales administration). They are not tied to the customer who triggered them. The result is an average that smooths everything out: simple customers subsidise complex ones, and nobody sees it.

Robert Kaplan and V.G. Narayanan’s work on customer profitability describes this with the “whale curve”: a minority of customers generates more than 100% of total profit, while some of the others erode it. Cost-to-serve is the tool that lets you build that curve for your own portfolio, from your own costs.

Beware of a common misreading: the goal is not to drop loss-making customers. It is to understand which behaviours (fragmented orders, rush requests, returns, payment delays) destroy value, then act on them through price, terms or organisation.

The 6 Cost-to-Serve Families

To stay pragmatic, we group service costs into six families. Each one matches an activity you can observe in the ERP and a cost driver, meaning the unit that makes the cost vary.

  1. Order intake and processing: entry, price and availability checks, sales administration, chasing order acknowledgements. Typical driver: number of orders or order lines.
  2. Picking and transport: picking, packing, palletising, freight, express or multi-address deliveries. Drivers: lines picked, parcels, pallets, kilometres or delivery points.
  3. Returns and credit notes: receipt, inspection, restocking or scrapping, credit note issue. Drivers: number of returns, value of credit notes.
  4. After-sales service and support: tickets, call-outs, warranties, billing disputes. Drivers: number of tickets, support hours.
  5. Receivables financing: the financial cost of the payment term, calculated on average outstanding balance and the company’s cost of funds. Driver: the customer’s DSO multiplied by their revenue. For managing payment terms, see our guide to accounts receivable, credit management and DSO reduction.
  6. Customisation and data exchange: custom labelling, dedicated packaging, EDI, supplier portals, bespoke reports, quality specifications. Drivers: number of active special requirements, EDI messages, set-up hours.

Alongside these six families sits one cross-cutting dimension: off-invoice discounts and rebates (year-end rebates, co-op marketing funds, goodwill gestures). They are not an activity cost, but they make the difference between list price and actual net price. They belong in the same view.

Cost Driver Table

FamilyMain driverGeneric ERP sourceWatch-out
OrderNumber of orders, linesSales documents (headers, lines)Separate manually keyed orders from automated EDI orders
Picking / transportLines picked, parcels, pallets, delivery pointsDeliveries, pick orders, shipmentsActual freight costs vs standard rates
ReturnsNumber and value of returnsCustomer returns, credit notesLink the credit note to the original sale
After-salesTickets, hoursService module or ticketing toolOften outside the ERP, reconcile by customer
FinancingDSO per customer, average outstanding balanceAccounts receivable, due dates, paymentsUse the real cost of funds
Customisation / EDIActive special requirements, messagesCustomer master data, exchange logsRequires a declarative inventory
DiscountsOff-invoice discounts, commercial credit notesCommercial terms, credit notesAccrue rebates as you go

A 5-Step Method

Step 1: Choose the Scope

Do not start with “all customers, all channels”. Pick a scope you can state in one sentence: for example “the top 150 customers in the B2B distribution channel”, or “customers of the UK subsidiary”. A useful first model covers most of your revenue plus the customers you suspect are the most expensive to serve.

Also decide on the reading axis: customer, channel (retail chains, distributors, e-commerce, direct sales) or segment. The ERP can support all three provided the customer record carries the right attributes (channel, segment, region) in its master data.

Step 2: Identify the Drivers

For each of the six families, select one driver, two at most. The classic trap is to multiply them: the model becomes unreadable and nobody will defend it in a sales meeting. A good driver meets three criteria: it can be measured in the ERP, operations teams understand it, and it genuinely varies from one customer to another.

Step 3: Collect the ERP Data

Extract driver volumes per customer over a rolling 12 months. Then gather the cost of each activity from management accounting: team payroll, transport costs, financial charges, equipment depreciation.

ERP Data Needed per Family

  • Order: sales order headers and lines, entry channel, creation and confirmation dates.
  • Picking / transport: delivery orders, lines picked, packing details, carrier, reconciled freight invoice.
  • Returns: customer returns, return reasons, linked credit notes and the original invoice reference.
  • After-sales: tickets or call-outs linked to the customer, time spent, parts consumed.
  • Financing: invoices, due dates, actual payment dates, outstanding balance per customer, cost of funds for the period.
  • Customisation / EDI: customer records (special requirements), message exchange logs, special price lists.

One finding recurs in almost every project: after-sales and customer-specific requirements often live outside the ERP (ticketing tool, spreadsheets, team memory). Plan to reconcile on customer ID from the start.

Step 4: Allocate the Costs

The cost of each activity is divided by the total volume of its driver to obtain a unit rate (cost per order, per line picked, per return). That rate is then multiplied by the volume each customer consumes.

Simplified ABC vs Time-Driven ABC

Two approaches coexist. For activity-based costing in the main ERP systems, our dedicated guide to activity-based costing covers the configuration.

CriterionSimplified ABCTime-driven ABC (TDABC)
PrincipleActivity cost divided by driver volumeCost per minute of capacity multiplied by the standard time of each operation
Data requiredCosts per activity, volumesCapacity cost, unit times per operation type
Set-up effortLow to mediumMedium to high
AccuracyGood if operations are homogeneousBetter when complexity varies widely between customers
MaintenanceUpdate rates every quarterUpdate standard times whenever a process changes
Recommended forFirst model, limited scopeHighly heterogeneous warehouses and services, advanced analytical maturity

Time-driven ABC was proposed by Robert Kaplan and Steven Anderson to avoid the heavy staff surveys of traditional ABC. Our recommendation: start with simplified ABC. Move to time-driven only if the gap between actual and allocated costs stays too wide on certain activities.

Step 5: Report the Results

A model that never leaves the controlling team changes nothing. Reporting must be short and actionable: a waterfall P&L per customer (revenue, discounts, gross margin, then each of the six families, finally net margin after cost-to-serve), a view by channel, and a ranking of the 20 most and least profitable customers. Present it to the sales committee with concrete cases, not just averages.

Worked Example (Fictional): Three Customers, Same Revenue, Opposite Profitability

Purely illustrative example. The customers, volumes and costs below are invented to show the mechanism. They do not reflect any real company. Assumptions: 30% gross margin on each, cost of funds of 5% per year.

Three customers each generate €1,000,000 in revenue, or €300,000 in gross margin.

Item (in €)Customer A: buying groupCustomer B: small distributorCustomer C: industrial
Revenue1,000,0001,000,0001,000,000
Gross margin (30%)300,000300,000300,000
Order3,00035,00012,000
Picking and transport22,000120,00060,000
Returns6,00055,00020,000
After-sales3,00060,00025,000
Financing (A: 30 days, B: 90 days, C: 60 days)4,10012,3008,200
Customisation / EDI6,00030,00010,000
Total cost-to-serve44,100312,300135,200
Net margin after cost-to-serve255,900 (25.6%)-12,300 (-1.2%)164,800 (16.5%)

Reading the table: customer A, with a few large orders, keeps most of its margin. Customer B, who multiplies small orders, returns and support requests, consumes more than its gross margin. In conventional reporting, A, B and C are three “30% customers”. In reality, one is funding the others.

The Decisions That Follow

Cost-to-serve is not an accounting exercise: it feeds five commercial levers.

  • Order thresholds. A minimum order value, or a free-shipping threshold, targets the fixed cost per order and per delivery directly. In the example, this is the main lever for customer B.
  • Shipping and service fees. Charging for urgent or split deliveries, or a handling fee for small orders, passes the driver back to whoever triggers it.
  • Payment terms. The financing cost per customer gives you a quantified argument to renegotiate a term, offer an early-payment discount or make a rebate conditional.
  • Catalogue rationalisation. Combined with margin by product (see our guide to gross margin by product line), cost-to-serve shows which items generate costly returns or picking.
  • Pricing. Differentiated prices by service model (collection, standard delivery, dedicated delivery) replace the single discount that hid very different costs.

Before any decision, validate the diagnosis with the sales team on two or three concrete customers. A quantified demonstration on an account they know well is more convincing than the overall model.

Limits and Pitfalls

  • Data quality. Duplicate customers, poorly populated channels or credit notes not linked to the original invoice skew the rates. A quick audit of customer master data is a prerequisite.
  • Sales resistance. Salespeople see cost-to-serve as a threat to their portfolio. Involve them from step 2 and present the model as a negotiation tool, not a sanction.
  • Over-complexity. Beyond a dozen drivers, the model becomes impossible to explain and maintain. A simple model updated every quarter beats an exhaustive one that is obsolete after six months.
  • Fixed-cost allocation. Cost-to-serve is a decision tool, not an accounting result. A warehouse cost allocated to a customer does not disappear if you lose that customer: measure the capacity freed up before drawing conclusions.
  • Commercial side effects. A price increase or a badly calibrated threshold can drive away a strategic customer. Simulate before applying.

Tooling: Native ERP, BI/EPM, Controlled Spreadsheet

OptionStrengthsLimitsRecommended maturity
Native ERP (analytics, management accounting)Data at source, traceability, no exportsMulti-activity allocation sometimes rigid, after-sales out of scopeCustomer analytical dimensions already in place
BI on a data warehouseCombines ERP, ticketing, transport and EDI; rich reportingModelling project, governance of definitionsData or BI team available
EPM / planningDriven allocation, simulation, pricing scenariosLicence cost, ERP integrationAdvanced management accounting
Controlled spreadsheetFast, inexpensive, good for a pilotRisk of errors, multiple versions, no traceabilityFirst pilot of 2 to 3 months

To understand where the ERP stops and the EPM layer begins, see our comparison of ERP and EPM integration: SAP, Oracle, Anaplan, OneStream.

Recommendation by maturity:

  • Beginner: a controlled spreadsheet (one data source, locked formulas, a single owner) on a limited scope, to validate the value.
  • Intermediate: move the model into the ERP’s analytics or a BI tool fed by the ERP, refreshed monthly.
  • Advanced: integrate with EPM to simulate the impact of changes in commercial terms before deciding them.

30-Day Starter Checklist

  1. Week 1: choose the scope (customers, channel, 12-month period), name a sponsor in sales leadership and a lead in management accounting.
  2. Week 1: confirm the six cost families and select one driver per family.
  3. Week 2: extract volumes per customer from the ERP (orders, lines, deliveries, returns, due dates) and check master data quality.
  4. Week 2: collect activity costs from accounting and from the warehouse, logistics and customer service managers.
  5. Week 3: calculate unit rates, build the customer P&L with simplified ABC, and spot the clearest gaps.
  6. Week 3: check two or three cases with operations teams to correct allocation errors.
  7. Week 4: present the results to the sales committee, choose one lever (order threshold, shipping fee or payment term) and a pilot customer to apply it to.

Going Further

Cost-to-serve is only worth the decisions it triggers. To frame the model, check your ERP data and turn the results into commercial terms, contact the DigitsLane team via the contact page: we will build a pilot with you on a limited scope. To keep reading, see also our guide to customer profitability analysis and our ERP and EPM comparison.