In a typical first-round profitability case, the candidate who ranks customers by revenue keeps the wrong ranking for the rest of the interview. The sheet rewards the candidate who notices that cost-to-serve is not proportional to sales, allocates it, and watches the leaderboard invert.
That is the thesis. Customer profitability cases are not a fancy profitability tree with a "customers" branch glued on. They are tests of whether you treat service activities as real costs that belong to named buyers, the way activity-based costing (ABC) has done since Cooper and Kaplan's work in the early 1990s. After this piece you should be able to run the allocation math, sketch a whale curve in two minutes, and recommend fix, reprice, or exit without sounding like you are firing the client's biggest logo.
What the interviewer is actually scoring
Firms describe the case as a client-style problem, not a memorized framework. McKinsey's interviewing page frames it as problem solving in a scenario that resembles client work. Bain's interviewing process asks you to work through an actual client problem with assumptions and quick math. BCG's case preparation guidance likewise emphasizes how you think through consulting-style problems.
On a customer profitability prompt, that translates into four moves on the scoring sheet:
- Separate product cost from customer-sustaining cost.
- Pick a cut that changes both revenue and service load (order pattern, channel, customization), not a cut that only redescribes the account list.
- Show the ranking after allocation, not before.
- Recommend an action that matches the economics: change the service model, change the price, or exit only when the relationship is structurally underwater.
I stop listening when someone says "enterprise customers are strategic" without a contribution number. Strategy language without cost-to-serve is decoration.
The ABC idea candidates underuse
Robin Cooper and Robert S. Kaplan's Profit Priorities from Activity-Based Costing (HBR, May–June 1991) is the primary source most prep material never opens. ABC separates unit, batch, product-sustaining, and facility costs, then extends the same logic to customers: start with contribution after product costs, then subtract customer-sustaining expenses such as sales calls, technical support, rush shipments, and collections effort.
Their Kanthal example is still the cleanest public story. Heating-wire maker Kanthal found that the casual 80/20 sales rule was wrong for profit. Roughly 20% of customers generated about 225% of profits; the middle ~70% hovered near break-even; and about 10% of customers destroyed roughly 125% of profits. Some of the largest-volume buyers were among the biggest loss-makers because they demanded low prices, frequent small deliveries, heavy technical support, and product changes. Management did not start by firing accounts. It changed prices, minimum order sizes, and how information flowed so that the same logos became contributors.
That pattern is what you are auditioning for in the room: diagnose the activity drivers, then repair the relationship before you recommend exit.
The whale curve is the picture of the same insight
Kaplan and V. G. Narayanan later summarized cumulative customer profitability as a "whale curve": customers ranked from most to least profitable, with cumulative profit plotted on the vertical axis. As summarized in standard references on customer profitability analysis, the top ~20% of customers often generate 150–300% of total profits; the middle 60–70% roughly break even; and the bottom 10–20% erase 50–200% of profits until the firm is left with 100% of realized profit. The gap between the peak of the curve and that 100% line is unrealized profit sitting inside the current book of business.
In an interview you do not need a polished chart. You need the sentence: "If we rank by contribution after cost-to-serve, the top fifth of accounts may be carrying more than the company's reported profit, and a thin tail is eating the rest."
Where cost-to-serve shows up in public numbers
True customer-level cost-to-serve is almost never published. Regulators and annual reports still leak the idea through segment or unit metrics.
EDF Energy's Consolidated Segmental Statement for 2023 allocates indirect costs (customer service, metering, bad debt, and related support) across domestic and non-domestic electricity and gas using activity drivers such as contact volumes and account volumes when costs cannot be assigned directly. That is ABC language in a statutory supply disclosure: contact intensity and account count drive cost-to-serve, not revenue alone.
BNY's 2024 annual report reports driving unit cost-to-serve down by roughly 5% per custody trade and about 15% per NAV in traditional fund services, with Securities Services pre-tax margin rising from 25% in 2023 to 29% in 2024. Public companies rarely name which clients are cheap to serve; they still talk about cost-to-serve as a managed unit cost that moves segment margin.
In a case, treat those as permission to invent plausible activity rates. The interviewer expects drivers (orders, rush lines, on-site visits, returns), not a fake ERP dump.
Worked example: Northline Carton flips the ranking
Invented company, invented numbers. Northline Carton sells corrugated packaging to three mid-market buyers. Leadership loves Apex Foods because Apex is 48% of revenue. Your job is to decide whether Apex is also 48% of profit.
Annual snapshot before cost-to-serve allocation (product cost already subtracted; figures in $000):
| Customer | Revenue | Gross margin after product cost | Share of revenue | Share of pre-CTS margin |
|---|---|---|---|---|
| Apex Foods | 4,800 | 1,440 | 48% | 46% |
| Harbor Brands | 3,200 | 1,120 | 32% | 36% |
| Clearpath Labs | 2,000 | 560 | 20% | 18% |
| Total | 10,000 | 3,120 | 100% | 100% |
On that table Apex looks like the prize account. Now allocate customer-sustaining cost using three activity pools Northline can actually count:
| Activity pool | Annual cost ($000) | Driver | Cost per driver |
|---|---|---|---|
| Order handling and picking | 720 | Orders | $120 / order |
| Rush / expedited freight | 360 | Rush lines | $400 / rush line |
| Field sales and engineering visits | 480 | On-site visits | $1,600 / visit |
| Total cost-to-serve | 1,560 | — | — |
Driver volumes by customer:
| Customer | Orders | Rush lines | Visits | Cost-to-serve ($000) |
|---|---|---|---|---|
| Apex Foods | 2,400 | 420 | 90 | 288 + 168 + 144 = 600 |
| Harbor Brands | 800 | 40 | 30 | 96 + 16 + 48 = 160 |
| Clearpath Labs | 2,800 | 440 | 180 | 336 + 176 + 288 = 800 |
| Total | 6,000 | 900 | 300 | 1,560 |
Contribution after cost-to-serve:
| Customer | Pre-CTS margin | Cost-to-serve | Contribution ($000) | Margin on revenue | Rank before | Rank after |
|---|---|---|---|---|---|---|
| Apex Foods | 1,440 | 600 | 840 | 17.5% | 1 | 2 |
| Harbor Brands | 1,120 | 160 | 960 | 30.0% | 2 | 1 |
| Clearpath Labs | 560 | 800 | -240 | -12.0% | 3 | 3 (loss) |
| Total | 3,120 | 1,560 | 1,560 | 15.6% | — | — |
Harbor, not Apex, is now the profit leader. Clearpath, a "growth" account with lots of small orders and site visits, is destroying $240k. Apex is still profitable, but it is no longer the account you protect at any price. That ranking flip is the entire case.
A miniature whale curve for these three: Harbor contributes 960, Apex adds 840 (cumulative 1,800), Clearpath subtracts 240 (ends at 1,560). Peak internal profit is 1,800 against realized 1,560; Clearpath is eating about 15% of the peak.
Current-period contribution is not the whole story. Retention, expansion, and acquisition cost belong in a lifetime-value frame; that sibling logic is covered separately when you need multi-year value rather than this year's economics.
Fix, reprice, or exit: the decision table
Once the ranking is honest, pick the action that matches the economics. Do not default to "fire the unprofitable customer."
| Situation | First move | Second move | Exit only if |
|---|---|---|---|
| High revenue, high cost-to-serve, still positive contribution (Apex) | Cap rush lines; introduce order minimums; tier service | Raise price on non-standard work | Contribution stays thin after two pricing cycles and capacity is scarce |
| Modest revenue, low cost-to-serve (Harbor) | Protect service levels; grow wallet share | Prefer Harbor when capacity is tight | Never exit for "smallness" alone |
| Negative contribution with repairable drivers (Clearpath if orders can batch) | Standardize SKUs; move to scheduled weekly drops; charge for visits | Reprice custom engineering | Drivers cannot change and strategic optionality is zero |
| Negative contribution, strategic logo, no path to repair | Isolate as a loss-leader with a board decision | Cap capacity given to it | You cannot name the strategic payoff in dollars or learning |
Ned's rule. Rank customers by contribution after cost-to-serve before you defend any logo. If your recommendation would be identical with and without the allocation, you have not solved a customer profitability case.
Practice this allocation under the clock
Run this drill cold, then score yourself:
- Open a profitability case and force a customer or channel cut within the first two minutes.
- Invent three activity drivers with unit costs; allocate; re-rank.
- Give a 45-second recommendation that uses the decision table above.
For a timed full case with synthesis feedback, start at /start (three typed turns, instant scores, no account required) or take the live voice case at /interview. Everything on CoachNed is open for seven days with no card; then $120 for a recruiting season or $49 a month. Unit-economics and break-even drills sit next door when the math layer is the leak.
Frequently asked questions
What is a customer profitability case interview testing?
It tests whether you can move from company-average profit to contribution by customer or segment after allocating cost-to-serve, then recommend fix, reprice, or exit. Revenue share alone is usually the wrong ranking.
How is cost-to-serve different from COGS in a case?
COGS (or product cost) follows units sold. Cost-to-serve follows activities: orders, rush lines, visits, returns, collections. Two customers with the same product mix can have very different cost-to-serve.
What is the whale curve in customer profitability?
It is a cumulative profit chart with customers ranked from most to least profitable. The peak often sits well above 100% of reported profit because loss-making accounts pull the curve back down.
Should I recommend firing unprofitable customers in the interview?
Usually not first. Interviewers prefer repair: minimum orders, service tiers, surcharges for rush work, or channel changes. Exit is for accounts that stay negative after realistic repairs and have no strategic value you can quantify.
How does this relate to customer lifetime value?
Current-period contribution after cost-to-serve is the base year. Lifetime value folds in retention, expansion, and acquisition cost over time. Use contribution for this year's ranking; use lifetime value when the prompt is about keep-versus-acquire or multi-year portfolio value.
Sources
- Profit Priorities from Activity-Based Costing (Cooper and Kaplan, HBR) — ABC hierarchy and the Kanthal customer-profitability pattern (roughly 20% of customers generating ~225% of profits). Checked 2026-09-24.
- Customer profitability analysis (overview of the whale curve) — Kaplan and Narayanan's whale-curve ranges (top ~20% often 150–300% of profits). Checked 2026-09-24.
- Measuring and Managing Customer Profitability (Kaplan and Narayanan, EconBiz) — bibliographic record for the 2001 Journal of Cost Management article on measuring customer profitability. Checked 2026-09-24.
- EDF Energy Consolidated Segmental Statement 2023 — public allocation of indirect supply costs using activity drivers such as contact and account volumes. Checked 2026-09-24.
- BNY 2024 Annual Report — disclosed unit cost-to-serve reductions (~5% per custody trade, ~15% per NAV) and Securities Services margin lift. Checked 2026-09-24.
- McKinsey interviewing — firm framing of case interviews as client-like problem solving. Checked 2026-09-24.
- Bain interviewing process — firm description of working through a client problem with assumptions and math. Checked 2026-09-24.
- BCG case interview preparation — firm guidance on thinking through consulting-style case problems. Checked 2026-09-24.
CoachNed is not affiliated with McKinsey, BCG, Bain, Deloitte, or any firm named above.