- Why the Standard P&L Cannot See This
- The Methodology: Five Steps From Blended Average to Plugged Leak
- The Fix-First Principle
- The Five Structural Leaks and How to Plug Them
- Executing the Exits
- What Plugging the Leaks Unlocks
- Customer Profitability Analysis: Operator FAQ
- About the Stagnation Assassin
The trailing gross margin smooths every leak into an acceptable average.
That sentence is the whole reason this page exists, so sit with it. A portfolio that aggregates to defensible numbers can contain individual customer-product combinations destroying value every single day they are served, and the standard income statement is structurally incapable of showing you which ones. Hypothetical: picture a business reporting a healthy 32 percent blended gross margin, respectable, on-plan, unremarkable in the operating review. Decompose that same portfolio to the customer-product level and the blended number splits into a spread running from 55 percent on the best combinations down to negative 12 percent on the worst, with the profitable core quietly subsidizing a tail of value destruction that has never once appeared on a report anyone reads.
Nobody built that tail on purpose. It accreted, one reasonable exception at a time, and the averaging machinery of standard accounting has been laundering it ever since. Customer profitability analysis is the method for un-laundering it: finding the leaks, sizing them, and plugging them, in a sequence that funds itself as it runs. In the transformation methodology this is Upgrade 1, Plug the Leaks, and it goes first for a structural reason: most of it requires nobody’s permission, and everything it recovers pays for the upgrades behind it.
Why the Standard P&L Cannot See This
Standard cost accounting spreads shared costs evenly or by revenue across the portfolio, while service intensity is wildly unequal across accounts. The allocation systematically flatters the expensive-to-serve and penalizes the easy-to-serve, and the blended gross margin averages every leak into an acceptable number. The standard P&L is not just missing the leaks. It is pointing you away from them.
Before the method, understand the blindness, because it is not carelessness. It is architecture.
Standard cost accounting allocates shared costs evenly, or by revenue, or by volume, across the portfolio. Order processing, customer service, freight coordination, engineering support, inventory carrying, planning time: spread like peanut butter. The allocation is defensible in an audit and catastrophic in a diagnosis, because service intensity is wildly unequal across accounts. The account that orders full truckloads on a predictable schedule, holds standard configurations, and pays in thirty days consumes a fraction of the shared-cost pool its revenue-based allocation charges it. The account that orders erratically, demands customization, generates expedites, and stretches payables consumes a multiple of its allocation. Peanut-butter costing systematically flatters the expensive-to-serve and penalizes the easy-to-serve, which means the standard P&L is not just missing the leaks. It is actively pointing you away from them.
The correction is cost-to-serve: attaching the true activity costs to the accounts and combinations that generate them. Done as a full activity-based-costing project, this takes a year and a consulting team. Done at the precision a banded diagnostic actually needs, it takes weeks, and the estimation shortcuts are catalogued in the companion guide Cost-to-Serve Analysis: The 9 Hidden Costs. The methodology below assumes the fast version, because the trajectory you are diagnosing does not pause for the elegant one.
The Methodology: Five Steps From Blended Average to Plugged Leak
Customer profitability analysis runs in five steps: build a customer-product matrix from twelve months of invoices, rank every combination by gross profit, drill recursively with the 80/20² method, layer in cost-to-serve, and segment every leaking combination into one of four treatments: fix, reprice, restructure, or exit.
Step One: Build the Customer-Product Matrix
Code every transaction over the trailing twelve months by customer and by product, then compute gross profit for each customer-product combination. Not customer totals. Not product totals. Combinations. This is the move most 80/20 analyses never make, and it is the move that matters, because the leaks live at the intersection: a customer who looks fine on the customer ranking, buying a product that looks fine on the product ranking, in a combination that loses money on every order. Single-axis analysis cannot see it by design. The combination axis is the only place it becomes visible.
The data demand is invoice-level history, which every ERP can produce, and the precision demand is directional, not forensic. If a combination’s computed margin, perturbed by your data uncertainty, would still land in the same decile, the data is good enough. Move.
Step Two: Run the First-Level 80/20
Rank all combinations by gross-profit contribution, descending, and find the 80 percent line. The result will not surprise you: roughly 20 percent of combinations produce roughly 80 percent of gross profit. Every operator has seen this chart. Most operators stop here, congratulate themselves on knowing their concentration, and file it. Stopping here is why the leaks survive.
Step Three: Drill Recursively. This Is the 80/20² Method
Ask the question the standard analysis never asks: within the top 20 percent, is there another 80/20? There is. Rank the first-level set internally, find the 80 percent mark again, and you have the 80/20² set: roughly the top 4 percent of the full portfolio. At the cart company, that was 41 combinations producing 64 percent of total gross profit. Drill once more and you have the 80/20³ set, roughly the top 0.8 percent, a handful of combinations carrying a staggering share of the entire company’s economics.
At the cart company, 41 combinations produced 64 percent of total gross profit, and the eight 80/20³ combinations were all top-customer, automated-line, standard-configuration carts. The structural core of the business was narrower than anyone running it believed.
The recursive drill produces a four-tier map, and each tier gets a different posture. The 80/20³ set is the structural core: the combinations the operation is truly built on. These get protected: capacity priority, service priority, defended terms. The 80/20² set is the strategic core: managed for growth and margin, the combinations the commercial organization should be pointed at. The remainder of the first-level 80/20 set is the supporting portfolio: served efficiently, standardized, watched. And everything outside the first-level set is the long tail, where the leaks live. It gets repriced, restructured, or exited, and the rest of this page is about how.
The full walkthrough of the tiering framework, with a 20-customer worked dataset, lives in the companion guide to The 80/20 Matrix of Profitability.
Step Four: Layer In Cost-to-Serve
Invoice-level gross margin understates the tail’s damage, because the tail’s damage is mostly in activity costs the invoice never sees. Attach the nine hidden costs, expedites, engineering time, dedicated inventory, payment-term financing, returns, call intensity, small-order transaction load, special packaging, meeting burden, using the estimation methods in the cost-to-serve catalog, and re-rank. Combinations that looked marginal go negative. Combinations that looked acceptable go marginal. The bottom decile, fully loaded, is almost always destroying value, and now you can prove it with numbers a CFO will sign.
Step Five: Segment Into Fix, Reprice, Restructure, Exit
Every leaking combination gets one of four treatments, and the sequence within the treatments is the difference between a surgical operation and a spreadsheet massacre. Which brings us to the principle that governs the whole prescription.
The Fix-First Principle
Exit is the last resort, not the first move. A large fraction of leaking combinations can be converted to positive contribution with a price adjustment, a minimum order quantity, a service-tier change, a logistics change, or a configuration substitution, keeping the revenue, the relationship, and the capacity absorption at restored economics.
Before you fire anyone, understand this: exit is the last resort, not the first move.
The spreadsheet-brutal version of this analysis, beloved by a certain kind of consultant, ranks the portfolio, draws a line, and fires everything below it. It feels decisive. It is lazy, and it burns value, because a large fraction of leaking combinations can be converted to positive contribution with a move short of exit: a price adjustment, a minimum order quantity, a service-tier change, a logistics change, a configuration substitution. A fixed combination keeps its revenue, keeps its relationship, and keeps its capacity absorption, at restored economics. An exited combination surrenders all three to recover the leak. Fix beats exit wherever fix is available, and fix is available far more often than the massacre crowd admits.
The Five Structural Leaks and How to Plug Them
Leaks have geography. They cluster in five places: the long-tail customer drain, the bottom-decile SKU drain, the manual-process exception, the customer-product mismatch, and the legacy contract drag. Each has a specific plug, and the execution sequence runs from unilateral moves this month to negotiated moves last, because sequencing is strategy.
Where the leaks cluster, and how each cluster gets plugged, is not random. Knowing where each leak hides means finding them in days instead of quarters.
Leak 1: the long-tail customer drain. Every B2B company carries a tail of small accounts that ordered once, ordered erratically, or ordered on terms negotiated by a salesperson who left years ago. Combined revenue: 3 to 8 percent of total, which is exactly why nobody examines it. Combined profit, fully loaded: frequently negative, because each small account consumes a baseline of setup, order processing, service, and inventory exposure wildly disproportionate to its revenue. The plug is tiered exit: bottom-decile accounts get a minimum-order-quantity notification, a price increase sufficient to reach positive contribution, or a clean exit with a referral to a competitor better positioned to serve them. Most companies fear the optics. The optics are almost always better than the fear, because long-tail customers are invisible to the broader market and their exit is rarely noticed outside the company.
Leak 2: the bottom-decile SKU drain. SKUs launched for a customer that no longer buys, kept in the catalog because someone might order one, or grandfathered from an acquisition nobody fully integrated. They consume engineering hours, carrying cost, setup time, and management attention out of all proportion to contribution. The plug is portfolio rationalization: discontinue what has not been ordered in twelve months, reprice what sells in single-digit quantities to recover carrying cost, and bundle or substitute what only the long tail buys. The portfolio shrinks 15 to 30 percent in most operations, and complexity drops by more than that. The product-axis version of this whole discipline is covered in the companion guide The Money-Losing SKU.
Leak 3: the manual-process exception. The class of orders that runs through the manual or semi-automatic process instead of the automated one, always justified by a commercial argument: the customer needs this configuration, the order protects the relationship, the volume is too small to automate. It is almost always structurally underpriced, because the pricing was set when the exception was rare and never adjusted as it grew, and it is almost always the largest single dollar leak in any operation running mixed automated and manual production. At the cart company, this leak was the 39 percent of top-customer volume cascading onto the manual line, the leak that started the entire methodology. The plug is structural, not commercial: expand automated capacity, add flexibility tooling, or shift the volume to a product the automated process can produce. Repricing the exception is the interim move; eliminating the exception is the fix.
Leak 4: the customer-product mismatch. A high-volume customer buying a low-volume configuration, or a low-margin customer buying a high-complexity product. Invisible to single-axis analysis by definition: the customer looks fine on the customer ranking, the product looks fine on the product ranking, and the combination is the leak. The plug is reconfiguration. Where a high-value customer is buying a low-fit product, develop or substitute one that fits the actual use case; the cart company’s worked example was discovering that a secondary grocery banner buying standard small carts actually wanted a bigger cart that did not yet exist in the line, and building it converted a mismatch into a higher-margin replacement. Where a low-margin customer is buying a high-complexity configuration, reprice to true cost or stop selling that configuration to that customer.
Leak 5: the legacy contract drag. Every company operating more than five years holds at least one contract negotiated under conditions that no longer apply: input costs risen, mix shifted, volumes changed, terms untouched. Legacy contracts look acceptable on trailing gross margin because volume amortizes the loss, but on true unit economics they run negative, and they do a second, quieter damage: they anchor every subsequent negotiation, because the customer uses the legacy terms as the reference point. The plug is renegotiation, not exit, because legacy contracts live with significant customers and unilateral exit is rarely on the table. Bring the contract forward, share the cost data transparently, and negotiate a price adjustment, a volume commitment, a mix shift, or a structural change that restores contribution. The cart company’s top-customer master agreement was exactly this leak, and its resolution came through restructuring the relationship around a product category with better economics for both sides.
Sequence the five by ease of execution, and the sequencing is strategy, not convenience. The long-tail drain and the SKU drain are unilateral: no external negotiation, act this month, build visible early wins and organizational credibility. The manual-process exception is structural, slower, and the largest dollar recovery. The mismatch requires product development or substitution. The legacy contract requires negotiating with a counterparty who has leverage, which is why it goes last: you want the credibility of four executed plugs behind you when you sit down at that table.
Executing the Exits
Some combinations survive every fix attempt and still leak. For those, exit is right, executed as its own discipline: confirm exit beats reprice, align internally before the external conversation, protect the transition, and treat pushback as data. An exit executed with respect costs one account. A careless one costs your market reputation.
Exit done well is a discipline of its own: the three tests that confirm exit beats reprice, the internal alignment before the external conversation, the transition plan that protects your market reputation, and the reprice-as-exit-ramp maneuver, a price that either makes the account profitable or makes it leave, with the non-negotiable requirement that you be genuinely willing to keep them at that price. The scripts, the pushback handling, and the capacity-backfill question all live in the companion guide How to Fire a Customer Without Burning the Relationship. The only thing that belongs on this page is the standard: an exit executed with respect costs you one account. An exit executed carelessly costs you the story the market tells about you.
One reading of the pushback matters enough to state here, because it changes decisions: when you move terms on a leaking account and the account pushes back, the pushback is data, and it comes in two kinds. Friction is the buyer testing whether you fold; hold the move. Information is the buyer revealing value, volume, or a strategic role your matrix did not capture; weigh it, and occasionally reverse. The moves went out reversible for exactly this reason.
What Plugging the Leaks Unlocks
The typical first pass finds 5 to 10 percent of revenue fully-loaded unprofitable. Plugging that slice raises Profit Velocity immediately while freeing capacity, service bandwidth, and management attention rather than consuming them, which is why this work goes first in the 90-day cadence and funds every upgrade behind it.
Run the arithmetic on why this upgrade goes first, because the arithmetic is the argument.
The typical first pass finds that somewhere between 5 and 10 percent of revenue is fully-loaded unprofitable. Exiting or repricing that slice does something no cost program can: it raises Profit Velocity immediately, because the retained portfolio’s blended margin steps up the moment the leaking combinations stop dragging it, and it does so while freeing capacity, service bandwidth, and management attention rather than consuming them. In a Growth Trap, this is the move that halts the compounding: the marginal-dollar decay that defines the trap runs through exactly the combinations this analysis surfaces, and plugging them turns the growth engine from an accelerant of decay back into an asset.
The typical first pass finds that somewhere between 5 and 10 percent of revenue is fully-loaded unprofitable. Exiting or repricing that slice raises Profit Velocity immediately, while freeing capacity, service bandwidth, and management attention rather than consuming them.
In the 90-day cadence, this work is the spine of Days 1 to 30, and deliberately so: the unilateral leaks clear the 70 percent confidence bar, reverse cheaply if a read comes back wrong, require no capital and no committee, and produce the early recovered dollars that fund everything after them. That early-wins logic shows up everywhere serious turnaround work is studied, including Harvard Business Review’s guide to turning around nearly anything. The calendar placement, the judgment calls, and what follows in Days 31 to 90 live in the companion 90-Day Business Turnaround Playbook. And the natural next move on the retained portfolio, converting chronic underpricing into structural margin, is Upgrade 2: the complete guide to B2B price increases.
The blended average has been covering for the tail for years. It will keep covering until an operator decodes it. Build the matrix. Run the drill. Find the geography. Plug the leaks in sequence. The 32 percent business with the negative-12 tail does not need more revenue. It needs about six weeks of this.
Plug the Leaks is Upgrade 1 of the five structural upgrades in Ten Minute Transformation (Koehler Books, February 2027), which carries the full 80/20² method, the five leaks with their complete plug protocols, and the cart-company worked case end to end.
Customer Profitability Analysis: Operator FAQ
The questions operators ask most about customer profitability analysis, answered in capsule form: what the method actually is, why the standard P&L hides the damage, how the 80/20² drill works, how much revenue is typically unprofitable, and whether firing customers is ever the right first move.
What is customer profitability analysis?
Customer profitability analysis decomposes blended margin to the customer-product combination level, attaches true cost-to-serve, and ranks every combination by fully-loaded contribution. It exposes the spread the income statement averages away, from strongly profitable core combinations down to combinations that lose money on every order they generate.
Why does the standard P&L hide unprofitable customers?
Because standard cost accounting allocates shared costs by revenue or volume while service intensity is wildly unequal. Erratic, customized, expedite-heavy accounts consume a multiple of their allocation; predictable full-truckload accounts consume a fraction. The allocation is defensible in an audit and catastrophic in a diagnosis.
What is the 80/20² method?
The 80/20² method reruns the 80/20 ranking inside the top 20 percent of customer-product combinations, isolating roughly the top 4 percent of the portfolio. A third pass, 80/20³, isolates roughly the top 0.8 percent: the structural core the operation is truly built on.
How much revenue is typically unprofitable?
A typical first pass finds that between 5 and 10 percent of revenue is unprofitable once the nine hidden costs are attached. That slice never appears on the standard P&L because blended gross margin averages the losses against the profitable core that subsidizes them.
Should you fire unprofitable customers first?
No. Fix beats exit wherever fix is available: a price adjustment, minimum order quantity, service-tier change, or configuration substitution restores economics while keeping revenue, relationship, and capacity absorption. Exit is reserved for combinations that survive every fix attempt and still leak.
About the Stagnation Assassin
Todd Hagopian is a Fortune 500 transformation executive who has generated $3B+ in shareholder value across Berkshire Hathaway, Illinois Tool Works, Whirlpool, and JBT Marel, where he serves as VP of Global Product Strategy. Known as The Stagnation Assassin, he is the author of two published books: The Unfair Advantage: Weaponizing the Hypomanic Toolbox and Stagnation Assassin: The Anti-Consultant Manifesto. His blog is published in 15+ languages and read by operators worldwide. Bring him to your stage via his speaking page or connect with him on LinkedIn.
Your blended margin has been covering for the tail long enough. Send me one line about your P&L and I will run a 15-minute 80/20 Portfolio Performance Audit against it: the matrix, the drill, and the first leak worth plugging. Book the audit here.