The Margin Leak

Consider a programme that, by every measure anyone bothered to look at, was a success. It was sold to a client the firm had wanted for two years, at a price the commercial director was pleased with, for a scope both sides believed they understood. It delivered on time. The client was happy enough to take a reference call. The team that ran it was proud of it, and said so, warmly, at the closing review. On the wall of green and amber lights that a people business keeps to reassure itself, every light was green.

Then the quarter closed, and the finance system did the one thing it is very good at. It told the truth about the money. The programme that had been sold at a comfortable margin had delivered almost none. Not a loss. Not a scandal. Nothing that would trip an alarm or start an investigation. Just a number, at the foot of a column, a great deal smaller than the number the firm thought it had sold. The margin everyone believed they had won had, somewhere between the signature and the sign-off, quietly gone.

Gone where? That is the question that stops the room, and it stops the room because nobody can answer it. There was no disaster to point at, no theft, no single bad call that a person could be held to. The client did not renege. The team did not fail. If you laid the whole programme out end to end and asked, of each choice along the way, "was that reasonable at the time," the answer would come back yes, yes, yes, all the way down. And yet at the end of that long chain of reasonable, defensible, individually trivial decisions, the margin was not there. It had not left through a hole. It had leaked, through a hundred small doors, each one left ajar by somebody doing what looked, in the moment, exactly like their job.

This is the most expensive thing that happens in a people business. It is also the thing the business is least equipped to see.

The question at the foot of the column

There is a question every commercial leader asks at month-end, and a second question underneath it that they almost never get to. The question on top is the one finance answers cleanly: did we make the margin we planned? The question underneath is the one no system answers at all: if we did not, where exactly did it go, and could we have stopped it while it was going?

The first question is a scoreboard. The second is a diagnosis, and the diagnosis is the one that matters, because you cannot treat a scoreboard. By the time the margin shows up as missing in the accounts, the programme is delivered, the people have rolled onto other work, the decisions that drained it are weeks or months old, and all that remains to be done is to explain the variance in a meeting and resolve, vaguely and sincerely, to do better next time. The money is not coming back. It was never going to come back from a report.

This chapter is about the second question. This is the chapter where the abstractions of the earlier ones, the decisionless stack, the broken loop, the gap between the deal and the done, stop being ideas and turn into money.

Where the margin actually goes

A margin leak is not one thing that you could name and guard against. It is a whole category of small, local decisions, made in every corner of the business, each of which trades away a little margin for something that felt, in the moment it was traded, more pressing than margin. Walk the common doors one at a time, the way you might walk the systems of the stack, and you will recognise every one of them from your own firm.

The first door is scope quietly expanding. It begins as a small favour. The client asks for a little extra, a bit outside what was written down, and the person on the ground says yes, because the relationship matters and it is only an hour and raising a change request over an hour would feel petty and slow. Then the client asks again, and yes is now the precedent. No change note is ever cut, because cutting one now would mean admitting that the last three should have been cut too. Multiplied across the length of an engagement, the favours become a second, unpaid engagement running silently alongside the paid one. And notice what the first yes actually was: a pricing decision, made by a person with no pricing authority, who could not see the running total of everything that had already been given away.

The second door is the wrong people staffed on the work. A programme is priced on the assumption of a particular mix of seniority, so many senior hours, so many junior, blended to a cost the margin can carry. Then reality staffs it with whoever was free. The expensive person ends up doing the cheap task, a partner-grade specialist formatting a deck because they had a light fortnight. Or, more quietly and more often, the cheap person is left to do the expensive task, a junior handed a workstream that needed grey hair and judgement, so it drifts, and then a senior has to be pulled in to rescue it, at senior cost, on top of the junior cost already spent. Either way, a scheduling decision made on the basis of who had capacity has silently overwritten a margin decision made on the basis of who was priced in.

The third door is escalations handled late. A problem that would have cost an hour to resolve in week two costs a fortnight to resolve in week eight, because no one raised it while it was still an hour. The cost of an escalation is very nearly a function of how late it is caught, and lateness, here, is a decision, even though it never feels like one. It is the decision not to flag the thing while it was small, made silently, by omission, by everyone who noticed the crack and assumed someone else would mention it. The margin does not leak when the problem occurs. It leaks in the days nobody spent on it.

The fourth door is renewals under-priced out of fear. The renewal comes round, and the account manager who owns it would rather hold the price, or shave it, than open the conversation that might put the logo at risk. The fear of losing the account gets priced into the renewal as a discount that no one ever formally authorised as a discount, because it does not look like one. The renewal is repriced downward not by any analysis of what the work now costs to serve, but by anxiety, and anxiety is a poor pricing model, because it always argues for the same answer.

The fifth door is delivery heroics that mask a broken plan. This is the most dangerous door, because it is disguised as its opposite. The engagement came in, the client is delighted, but it only came in because one person worked three weekends and personally held three fragile things together through force of will. Everyone calls that a success. It is a margin leak wearing a medal. The heroics conceal the fact that the plan was wrong, that the work cost far more effort than it was priced to cost, and, worst of all, that none of it can be repeated or scaled, because it depended on a specific exhausted human being who will not always be available. The next time the same plan is sold, there is no hero free, and the plan simply fails, or the firm burns out the one person who could rescue it and loses them to a competitor who offers a quieter life. And there is a further cost, quieter and worse than all of these: the heroics poison the next plan. Because the programme came in, its apparent productivity becomes the benchmark, and the next engagement like it is priced to the same standard, as if that output were normal and repeatable, when in truth it was bought with one person's unpaid weekends. It sells, in good faith, a margin that only ever existed because someone was quietly destroying themselves to produce it.

The sixth door is the discount granted in the sale that delivery must silently absorb. The deal was won on a price that assumed an efficiency delivery cannot actually achieve, or a scope tighter than the one the client thinks they bought. The discount is banked in the CRM as a closed win, a good day, a bell rung. The cost of honouring it arrives months later, in a different system, on the desk of a delivery team that was never in the room when the promise was made and has no way of knowing which corners of the price were real and which were hope. This door lives exactly on the seam between the promise and the work. It is the workforce starting in sales, from Chapter 10, now showing up in the ledger: a promise made by someone who would never staff or deliver it, quietly setting the margin before the first hour of the work was ever planned.

The seventh door is hiring against assumptions that were already wrong. A people business hires ahead of demand it can only forecast, and the forecast is itself a set of decisions about which pipeline is real and which is hope. Hire too slowly and the work arrives with no one to do it, so it is covered by expensive contractors or an agency placement at a premium, or delivered late at a cost to the relationship. Hire too fast, or hire the wrong profile, and the firm carries people on the bench at full salary against work that never came, or staffs what it did win with whoever it happened to hire rather than whoever it needed. Reach outside for a skill that already existed one team over, because no one could see it, and you pay the premium twice: once in the agency fee, and once in the internal person who leaves because they were passed over for a job they could have done. Every one of these is a hiring decision made against a promise, the pipeline's implied demand, and when the promise was wrong the margin quietly pays the difference. Of all the doors, this is the one that reaches furthest across the firm, joining the sale that implied the demand, the recruiting that answered it, the delivery that lived with the answer, and the margin that absorbed the gap.

Look at the seven together and the shared shape is unmistakable. Every one of them is a decision. Every one is small enough to be beneath notice on its own. Every one is defensible, even admirable, in the moment it is made: a happier client, a filled slot, an avoided fight, a rescued deadline, a closed deal, a role filled. And every one is made by a person who cannot see, and is not accountable for, the aggregate consequence of it. Margin, in a people business, is everyone's to spend and no one's to protect.

Why the leak is invisible

If each of these is a decision that somebody made, and made knowingly, why can no one see the leak while it runs? There are three reasons, and they stack.

The first is that each decision is small. Absorbing an hour of scope, staffing the analyst who happened to be free, holding a renewal price to avoid a hard call: none of these, on its own, is worth a meeting. None trips a threshold. None is large enough to feel like the kind of thing a firm loses money on. The leak is invisible for the same reason a dripping tap is invisible on the water bill: no single drop is a problem, and nobody watches drops.

The second is that each decision is local. It is made inside one engagement, by one person, in one system or in no system at all: a yes in an email, a name typed into a resourcing sheet, a task quietly widened in a project tool, a price agreed on a call. It never rises to a level or a place where anyone holding a view of the whole could weigh it against the margin it was spending.

The third reason is the deep one, and it is the reason the first two never get corrected. The aggregate is never assembled anywhere. This is the decisionless stack, met again in its most costly form. Each system in the estate holds one facet of the leak and only one. The CRM holds the discount, as a closed-won figure with a cheerful green tick. The resourcing system holds the staffing, as a set of bookings. The project tool holds the scope, as a task list that grew. The finance system holds the result, and only the result, and only at the end. No system holds the sum, because the sum is not a record of a thing. It is a pattern across decisions, and the decisions were never captured.

The leak does not live in any of your systems. It lives in the seams between them.

And the seams are precisely where the people-business loop breaks. The gap between the deal and the done is not empty space; it is the busiest, most consequential territory in the firm, and it is the one stretch of ground that no system owns. The discount is decided in the sale and paid for in delivery, and the sale and the delivery live in different systems, are run by different people, and are measured on different numbers. The margin leaks across that seam, and a leak across a seam is invisible to any instrument that watches only one side of it. This is also why the warehouse, for all its power, cannot show you the leak either: it aggregates the records from both sides with great precision, but the leak is not made of records. It is made of decisions, and the decisions happened in the seam, and the seam is exactly what the warehouse joins over without ever seeing.

So the aggregate does get assembled, in the end, but only in one place and only once: in the finance system, at month-end, after the fact, as a variance with no explanation attached. The single instrument that finally sees the whole leak is the one instrument that can only see it once it is far too late to stop.

Revenue leaks and margin leaks

People businesses are not careless about money. Most of them are intensely, professionally vigilant about it. The trouble is that they aim almost all of that vigilance at the wrong leak.

There are two kinds of leak, and they are easy to confuse and expensive to conflate. A revenue leak is money you failed to win or failed to bill. It is the deal you lost, the hours you worked and never invoiced, the renewal that quietly lapsed, the scope you delivered and forgot to charge for, the invoice that went out late, or short, or not at all. A margin leak is a different animal. It is money you did win, and did bill, and did collect, and then gave away again in the delivery. The revenue arrived. You simply spent more earning it than you meant to, through the hundred small doors of the section above.

Now ask why firms obsess over the first and are blind to the second, because the asymmetry is not an accident. It is structural.

A revenue leak is visible because it shows up against a target. You set a number, you missed it, and the gap between them has a name and, usually, an owner. Sales guards the top line ferociously, because the top line is what sales is measured on. Billing chases every unbilled hour, because unbilled hours are its entire reason to exist. There are whole functions, revenue operations, credit control, commercial desks, whose standing job is to stop revenue from leaking. The revenue leak, in other words, has enemies.

The margin leak has almost none. It does not show up against a target, because no one ever set a margin target for the specific decision to absorb an hour of scope, or to staff the senior on the junior task, or to hold the renewal price out of nerves. Plenty of people will tell you that guarding margin is part of their job, and they are not wrong to claim it: the delivery lead, the project office, the commercial manager, the delivery-excellence team all watch some slice of it. But no one of them has a complete, live view across all the decisions that are spending the margin, because those decisions are scattered, held in different systems and different hands. Sales has left the field at the signing. Finance does not arrive until the close. In between, across the entire length of the delivery, no one holds the whole of it in view while it goes.

A revenue leak is money you never got. A margin leak is money you earned and then quietly handed back.

The margin leak ought to sting the more of the two. You paid to acquire that client. You paid, in proposal effort and discounting, to win the deal. You paid, in salaries and time, to deliver the work. And then, at the very last mile, having borne every cost of getting there, you leaked the reward. The revenue you never won was money you never had. The margin you leaked was money that was, briefly, genuinely yours, and you gave it back without noticing. Yet it is the one firms cannot see, because unlike a lost deal it never appears as a gap against a plan.

In the language of Chapter 11, the revenue leak is a failure of Win Confidence, and firms watch Win Confidence relentlessly, because the whole apparatus of sales is built to. The margin leak is a failure of Delivery Confidence, and almost no firm can read Delivery Confidence at all.

The instrument that catches it in flight

If the margin leaks because no one is watching delivery while the margin is being spent, then the cure is not a sharper post-mortem. You cannot fix a leak by getting better at describing the puddle. The cure is an instrument that watches delivery while delivery is happening, and that instrument, built properly, is the Delivery Confidence we met in Chapter 11.

Recall what it is. Delivery Confidence is the live, evidence-backed answer to a single question: can we still deliver what we sold, at the quality and the cost we promised? Not the status-deck version of that answer, the one where a project lead colours the box amber on a Friday according to how the week felt. A real, current reading of the state of each engagement, measured against the promise it was actually sold on.

That last phrase is the one that turns it from a report into a leak detector: against the promise it was actually sold on. A margin leak, stripped to its mechanism, is nothing more than a growing distance between what was sold and what it is costing to deliver. Delivery Confidence, done properly, holds both of those figures next to each other and watches the distance between them change. It carries the promise across the seam, the price, the scope, the assumed mix of seniority, the discount and what it assumed, out of the sale and into the delivery, so that the delivery team knows precisely what it inherited rather than reconstructing it from a contract. And then it reads the live state of the work against that inherited promise, continuously, not monthly.

Walk it back through the seven doors, and you can watch the instrument catch each one while it is still in flight. Scope drift becomes visible the third time the small favour is granted, not at month-end, because the work being done is being read against the scope that was sold, and the gap widens in view. The staffing mismatch surfaces at the moment of staffing, because the mix being booked is checked against the mix that was priced. The escalation is caught while it is still small, because the instrument is watching for the divergence rather than waiting for the client to complain. The renewal is priced against the true, now-visible cost to serve, rather than against fear, because fear loses its monopoly the instant the real number is on the table. The heroics show up honestly, as a plan running hot, a warning rather than a triumph, so the broken plan gets fixed and repriced instead of medalled and repeated. The sale discount is inherited by delivery as a known constraint, rather than a silent tax they discover only by missing their margin. And the hiring bet is placed against a pipeline whose reality is now visible, so the firm hires to demand it can actually see rather than to a forecast it merely hopes.

Delivery Confidence assembles the aggregate continuously, in the one place the decisionless stack never did, while the decisions are still being made, and therefore while they can still be made differently. It moves the discovery of the leak from the finance system at month-end back to the delivery of the work in real time. It converts a post-mortem into a control.

It is worth being precise about what the instrument does and does not do, because the fear here is that it becomes surveillance. It does not make the calls. It makes the calls visible, to the people who should be making them, in time to make them another way. The person granting the small favour can still grant it. They simply grant it now while looking at the meter, with the running total in view, so that a defensible local choice can be weighed against the aggregate cost it was blindly adding to. This is orchestration, not automation: the point is not to take the judgement away from the human on the ground, but to give that judgement the one thing the stack always denied it, a sight of the whole. For the first time, margin has a guardian while the work is live.

"We review project profitability every month"

Here is the objection that every competent finance leader will raise, and it deserves a straight answer, because the practice behind it is a good one. "We review project profitability every month," they say. "We sit down, engagement by engagement, and look at what each one is making. If margin were leaking, we would see it."

Take it seriously, because monthly profitability review is genuinely valuable and most firms should do more of it, not less. But it does not catch the margin leak, and the reasons it does not are structural.

The first reason is that a monthly review is a post-mortem on a wound that was live for thirty days. By the time the shortfall appears in the month-end numbers, the decisions that drained it are already weeks old. The review, held sometime in the following month, is examining the body and asking, thoughtfully, how it died. That is worth knowing. It is not remotely the same as keeping the patient alive.

The second reason is that you cannot manage a live process with a rear-view report. Delivery is a live process. Margin is not spent in a single event that a monthly cadence could catch; it is spent continuously, a little every day, in the ordinary flow of the work. Trying to control a continuous leak with a monthly snapshot is like trying to steer a moving car using a photograph of where it was, developed and delivered to you once every thirty days.

You cannot manage a live process with a rear-view report.

The third reason is the most damning, and it is the decisionless stack returning for its final appearance in this chapter. The review shows you that margin fell. It rarely shows you which decisions drained it. The monthly review is built on the records, the finance system's costs and revenues, and the records, as we established long ago, hold what happened and not what was decided. So the review tells you the programme made less than planned. The variance analysis narrows it to labour cost. And then it stops, because the thing you actually need next, the chain of small human decisions that put that labour cost there, was never captured in any system to be reviewed. You leave the meeting knowing the margin fell and not knowing which doors it left through, which means you cannot close them, which means next month the margin leaks out through exactly the same ones, and you hold exactly the same review about it.

A monthly profitability review is an excellent instrument for the question "did we make the margin we planned?" It is the wrong instrument, in principle and not merely in tuning, for the question "where is our margin leaking right now, and which decision do I change today to stop it?" The first question is accounting, and accounting runs at the speed of the ledger. The second is Delivery Confidence, and it has to run at the speed of the work.

The bridge

Something changes in a firm the moment it can see its margin drain while the draining is still happening. The leak stops being weather, the thing you accept about running a business made of people, and becomes what it always was: a set of decisions, made in the open, that can be made differently. A firm that can watch its own delivery as a live system, measured honestly against the promises it sold, has done more than plug a leak. It has started to operate in a fundamentally different way.

And that is the larger thing this chapter has been quietly building towards. Watch delivery in flight, and you will want to watch the sale the same way, and the renewal, and the learning that ought to flow from one engagement into the pricing of the next. Connect the three confidences instead of guarding three separate numbers in three separate rooms, and you no longer have a set of departments each defending its own figure while the margin leaks through the seams between them. You have one connected way of running the whole firm. That is the subject of the next chapter. Put the loop, the confidences, and the decision room together, and what you get is not a better report. It is an operating system for a people business: a firm built, deliberately, so that it does not leak.