Written to Win, Not to Keep
An estimate that is never checked is an estimate that never improves.
Executive Summary
The benefit case is the most carefully written document in the life of an investment, and it is read most closely on exactly one day — the day the money is decided. After that, it is rarely opened again with the same attention. This essay asks why the benefits that justify our programmes are so reliably forgotten once those programmes are funded, and argues that the answer is not carelessness but structure. The machinery of investment appraisal concentrates almost all of its rigour at the front gate and almost none at the back; the ownership of a benefit passes, silently, at the moment of handover, from a sponsor who promised it to an operation that never agreed to it; and the competition for scarce capital quietly rewards the most optimistic case rather than the most honest one.
I take seriously the case for the defence — that a business case is a decision instrument and not a contract, and that pursuing every last benefit into the management accounts costs more than it will ever recover — and conclude that the objection proves too much. An estimate that is never checked is an estimate that never improves; a forecasting discipline with no feedback loop does not stay honest, it decays. The correction is less a technique than a temperament, and it belongs at the level of the portfolio rather than the single project — because only the portfolio can see the same saving counted three times, and only the portfolio lives long enough to be held to its word.
The High-Water Mark
Picture the meeting on the morning a large business case is approved. The room has read the paper, or enough of it. The financials sit on the third page: a net present value that clears the hurdle rate, a payback comfortably inside three years, a benefits total large enough to make the capital ask look modest beside it. There is a short discussion, usually about cost and only rarely about benefit, and then the decision is minuted. The sponsor gathers the folder and leaves with what they came for.
That morning is the high-water mark of a benefit’s life. Everything after it runs downhill. The benefits section — the pages that did the real work of winning the money — will, in most organisations, never again be read with the attention it received that day. It has done its job. It secured the funding. Whether it still describes anything that will actually happen becomes, from the moment the vote is taken, very nearly irrelevant to everyone in the room.
I have sat through enough of these approvals to notice a small, telling ritual of grammar. Before the decision, the benefits are spoken about in the future perfect — the programme will have delivered twelve per cent, the function will have been consolidated. Once approved, they slip out of tense altogether. They stop being claims about a future that must be made to happen and settle into a fact of the record. The number that was a forecast on Wednesday is a commitment on Thursday and a memory by the following quarter.
An Instrument of Persuasion
We tell ourselves that a benefit case is a forecast. It is more honest to admit that, in the moment that matters, it is an instrument of persuasion. Its purpose is not to predict but to prevail — to clear a particular hurdle, in open competition with other papers doing exactly the same thing, before a committee with far more demands on it than capital to meet them.
Once we see the document for what it is, its later neglect stops being a mystery. Nobody maintains a persuasive instrument after it has persuaded. An advocate does not return to a winning argument to check whether it was, on reflection, entirely fair. The benefit case is written to be won with, and a thing built to win an argument is naturally abandoned the moment the argument is over.
A benefit case is written to clear a gate, not to be lived with. We should not be surprised that a document built to win an argument is set aside the moment the argument is won.
The trouble is that we then file the persuasive instrument as though it were an operating plan, and we are quietly astonished, two years on, to discover that the plan was never followed — because it was never really a plan. It was a bid. And a bid, having succeeded, has no further use. This is the first and most human of the structural forces at work: the document that justifies the spend and the document that would govern the delivery are, in most organisations, the same document, asked to do two jobs that pull in opposite directions. Persuasion wants confidence and a clean total. Governance wants doubt, ranges, and named owners. The persuasive job comes first, wins, and having won, crowds the other out.
The Asymmetry of the Gate
Consider, without flinching, where our disciplines actually sit along the life of an investment. At the front we have built an elaborate and genuinely impressive apparatus. There is a mandated appraisal template. There is discounted cash flow, scrutinised line by line. There is, increasingly, a gateway review before the money is released, and a sponsor — a senior responsible owner, in the language many of us have adopted — whose name is on the case. There is an investment committee whose entire function is to say no often enough that the yes means something. Every one of these mechanisms exists to sharpen the decision to spend.
Now look at the other end, after delivery. What stands there? In most organisations, almost nothing. A post-implementation review that is optional in practice whatever the policy says on paper, scheduled for a date by which the team has dispersed, and honoured far more often in the breach than the observance. There is no gateway after delivery to match the one before it. There is no committee whose job is to ask whether the last round of yeses was earned. The asymmetry is not an accident of any single organisation; it is the shape of the discipline itself.
| Front of the investment | Back of the investment |
|---|---|
| A mandated appraisal template | An optional post-implementation review |
| Discounted cash flow, challenged line by line | A benefits total no one revisits |
| A gateway review before the money is released | No gateway once delivery is done |
| A named sponsor to make the case | No named owner to answer for it |
| Competition that sharpens every claim | No audience left to sharpen anything |
The consequence is a system exquisitely tuned to decide whether to spend and almost blind to whether spending worked. We invest heavily in the quality of the promise and nothing in the honouring of it. And because the back gate is missing, the front gate slowly loses its own discipline — a point I will come back to, because it is the hinge on which the whole argument turns.
The Benefit Nobody Owns
The deepest structural reason benefits are forgotten is that, at the moment of delivery, they quietly change hands — and no one ever shakes on it.
The sponsor promised the benefit. But the sponsor does not, as a rule, realise it. A programme delivers a capability — a consolidated system, a new process, a restructured function — and then closes. The benefit, if it is to appear at all, appears afterwards, in the running of the business, on someone else’s watch. That someone is an operational manager who was not in the room when the case was made, did not write the twelve per cent, and has inherited a target they never agreed to alongside the day job they already had. Ownership does not transfer at handover so much as evaporate. The person who promised has gone; the person who could deliver never signed up.
Let me make this concrete, in the way these things actually go. A back-office consolidation case promises four-point-two million over three years, built almost entirely on a single line: a twelve per cent productivity improvement across roughly three hundred finance and administration staff. The programme delivers. The system goes live, the processes are redesigned, the twelve per cent of freed capacity is, by any fair measure, genuinely created. And then nothing removes it. The freed hours are real, but freed hours are not a saving; they are an opportunity to make a saving that somebody must actually take. No one holds the posts open for deletion. The capacity is quietly reabsorbed into rework, into a backlog that had been building, into the ordinary tendency of work to expand into the time available. Two years later the function is the same size it always was, doing more, and the four-point-two million exists only in a folder. Nothing failed. The system works. The benefit is simply nobody’s to collect.
This is the pattern that recurs across benefit after benefit: the programme delivers the means of the benefit and books the benefit itself, as though a capability created were a value realised. They are not the same thing, and the whole distance between transformation intent and transformation reality lives in the gap between them.
The Optimism That Wins
There is a further force, and it is the least comfortable to admit because it implicates the appraisal process we are most proud of. Competition for capital does not select for the most accurate case. It selects for the most attractive one.
When a dozen papers compete for a fixed pot, the paper with the boldest defensible benefit and the leanest defensible cost wins. The sponsor who has honestly written a range — who says the productivity gain might be six per cent or might be fourteen and here are the assumptions — is competing against a sponsor who has written twelve as a flat number and moved on. Rewarded over enough cycles, this teaches everyone the lesson: confidence is funded, candour is not. We should not then be shocked that our forecasts run hot. We have built a machine that pays for optimism.
The public appraisal world has at least named this. The optimism bias that reviewers began insisting we correct for — the well-documented tendency of appraisers to underestimate cost and duration and overstate benefit — is now met, in the more disciplined shops, with an explicit uplift applied to the raw figures precisely because the raw figures cannot be trusted to be neutral. That is a mature response. But it treats the symptom. The disease is that nothing downstream ever confronts the estimator with the outturn, so the bias is never personally, professionally corrected. The uplift is applied by a reviewer; it is never learned by the author.
And optimism has a portfolio-level cousin that a single business case can never see: double counting. Consider three programmes running in the same portfolio, each with a benefit case that claims savings from the same shared finance function of forty posts. Read on its own, each case is reasonable. Summed across the portfolio, they promise to remove one hundred and twenty posts from a function that contains forty. No individual gate can catch this, because each case clears its gate alone. Only something standing above all three — something with a view of the whole book of promises — can notice that the organisation has sold the same forty posts three times over. Most organisations have no such vantage point, which is why the sum of their approved benefit cases, if anyone ever added it up, would describe a company far leaner and richer than the one that actually exists.
The Honest Objection
The strongest reply to all of this is not a defence of neglect; it is a defence of proportion, and it deserves to be met at full strength.
A business case, the objection runs, is a decision instrument, not a contract. Its real value is delivered entirely at the front: it forces the thinking, disciplines the sponsor into articulating why the money should be spent, and gives the committee a basis for choosing between competing claims on capital. Once it has done that, its work is genuinely finished. To then chase every benefit into the management accounts — to stand up a benefits-tracking function, to re-baseline, to reconcile promised savings against actual headcount quarter after quarter — is to build an expensive bureaucracy whose running cost may well exceed anything it recovers. Organisations are not short of things to measure. The honest position, on this view, is to accept that benefit cases are forecasts made under uncertainty, that many will be wrong, and that the mature response is to keep the thinking at the front sharp and not to squander effort auditing the past.
There is real truth in this, and any argument that pretends otherwise is not worth making. Benefits tracking absolutely can become a make-work industry that measures everything and improves nothing. Some benefits genuinely cannot be isolated from the ordinary noise of the business, and insisting on a number where none can honestly be had corrupts the discipline as surely as ignoring benefits altogether.
But the objection proves too much. Its own logic — that the value is in the quality of the thinking at the front — is precisely what the absence of a back gate destroys. An estimate that is never checked is an estimate that never improves. If no outturn ever returns to the estimator, the front-gate thinking we are being urged to protect is exactly the thing that quietly rots, because nothing corrects it. The forecasting does not stay honest on its own; it drifts, cycle after cycle, in the direction that gets funded. The feedback loop is not bureaucratic overhead bolted on to the discipline. It is the discipline. Remove it and you do not get a lean process that thinks well at the front; you get a process whose front-end thinking has no reason to be accurate and every incentive not to be. The choice is not between a heavy audit and a light one. It is between a forecasting system that learns and one that cannot.
Holding the Promise
If the objection is right that heavy benefit accounting is often self-defeating, and I think it is, then the correction cannot be more of the same machinery pointed backwards. It has to be a different and lighter thing, and it has to sit at a different level.
The unit of benefits realisation is the portfolio, not the project. This is the single most important structural move, and it follows directly from everything above. Only at portfolio level can the double-counted forty posts be seen for what they are. Only the portfolio persists — projects close, sponsors move on, but the portfolio is a standing thing that can hold a running account of what was promised against what was delivered, across years and across the careers of the people who made the promises. A benefit orphaned by a closing programme has somewhere to be adopted. A pattern of optimism in one sponsor’s cases, invisible within any single approval, becomes visible across a book of them.
Within that, a few disciplines do most of the work, and none of them is expensive:
- Capture the baseline before, never after. A benefit is the distance between a before and an after, and if the before was never measured, the benefit can never be proven and will always be arguable away. The cheapest possible moment to secure a baseline is before the change; the most expensive is never.
- Give every benefit an owner who will still be there. Not the sponsor who will have moved on, but the operational manager who will run the changed function — and secure their agreement to the target while they can still shape it, not after handover when it lands on them as an imposition.
- Refuse to book a number you cannot measure. A benefit that genuinely cannot be isolated should be stated honestly as a rationale, not smuggled into the total as a figure. A smaller, defended benefits total is worth more than a large one nobody believes.
- Build the back gate. A single review, at portfolio level, some months after delivery, that asks one question of each closed case: what did we promise, and what did we get? Not to punish the gap, but to feed it back into the next round of appraisal so the estimating slowly improves.
None of this is a tracking industry. It is a handful of habits, most of which cost less than the appraisal effort we already spend without a second thought at the front. The asymmetry of the gate is not corrected by making the back as heavy as the front. It is corrected by making the back exist.
The Temperament of Realisation
What all of this finally asks for is not a technique but a temperament — the willingness to return to our own promises after the excitement of making them has passed, and to be told by the outturn that we were wrong.
That is harder than it sounds, because everything about the way we fund change is arranged to spare us the return. The approval is a celebration; the reckoning, if it comes at all, arrives long after and lands on someone else. The sponsor is rewarded for the promise and rarely present for its keeping. To close that gap is to accept a kind of accountability the current system carefully avoids: to sign a number knowing that, two years on, someone will hold it up beside what actually happened and ask you to account for the difference.
“Benefits realisation is not, in the main, a technique we lack. It is a discipline we decline.”
We know how to write a benefit. We know, if we are honest, roughly how far our benefits tend to fall short. What we mostly avoid is the modest, uncomfortable machinery that would make us look — the baseline captured in time, the owner who signed up, the single review that closes the loop. We avoid it not because it is difficult or costly, though we tell ourselves it is both, but because it would replace the clean satisfaction of the approved case with the messier truth of the realised one.
Go back to that meeting on the Thursday morning, the folder closing, the sponsor leaving with the money. Nothing about that scene needs to change except one thing: the quiet knowledge, shared by everyone in the room, that the folder will be opened again — that the promise made this morning will, in time, be read back to the person who made it. That single expectation, more than any framework, is what turns a benefit from an instrument of persuasion into something an organisation actually intends to keep. Until we build it, we will go on writing benefits to win, and forgetting them the moment they have won for us — and the gap between what our transformations intend and what they realise will remain exactly the size of our unwillingness to go back and look.