When Benefits Are Political — Why the Approval Number Was Never a Forecast
The benefits figure in a business case is frequently not a forecast. It is a political instrument.
The Number That Was Never Meant to Be True
Every experienced practitioner learns, sooner or later, to read a business case in two registers at once. There is the case as written — the confident projection of savings, the payback period, the net benefit that comfortably exceeds the cost. And there is the case as understood by everyone in the approval meeting, which is a rather different document: a negotiating position, a bid for scarce capital, a set of numbers assembled less to describe the future than to clear a threshold.
I have come to believe that we do the profession a disservice by pretending these are the same document. The benefits figure in a business case is frequently not a forecast. It is a political instrument. And until we are honest about that, every attempt to improve the accuracy of our estimates is aimed at the wrong problem.
Why the Numbers Inflate
It is tempting to frame benefits inflation as dishonesty, but that framing is both unfair and unhelpful. The people producing these numbers are, for the most part, not lying. They are responding, rationally, to a system that rewards optimism at the point of approval and forgets it everywhere else.
- Approval is a competition, not an assessment. Business cases do not stand alone; they compete for a fixed pot against other business cases, each authored by someone with the same incentive to look attractive. A realistic case placed next to optimistic rivals does not read as prudent. It reads as weak.
- The estimator and the deliverer are rarely the same person. The figure is produced early, by people who will have moved on before anyone checks it. The cost of an inflated benefit falls on a future team, and future teams do not sit in the approval meeting.
- There is no penalty for being wrong later. Because the post-approval scrutiny of benefits is so weak — often absent entirely — an inflated number carries almost no downside. The sanction for overpromising arrives, if it arrives at all, long after it could change anyone’s behaviour.
- Sponsors want the project, not the estimate. A committed sponsor is not a neutral party seeking the truest number. They are an advocate who has already decided the initiative should proceed, and the benefits case is the instrument through which that conviction is expressed.
Set these forces alongside each other and inflation is not an aberration. It is the equilibrium. The system is, in effect, tuned to produce optimistic numbers, and it would be surprising if it produced anything else.
We keep trying to solve a problem of incentives with better technique. But no estimation method, however rigorous, survives contact with a process that rewards the person who returns the most attractive number and never troubles the person who returns an untrue one.
What the Textbooks Leave Out
The guidance on this is not wrong, exactly; it is just curiously silent on the part that matters. It tells us to base estimates on evidence, to adjust for optimism, to challenge assumptions, to seek independent review. All sensible. All aimed at the quality of the estimate.
What the guidance almost never says is that the estimate is doing a job, and that job is not primarily to be accurate. The number exists to move a decision. As long as that is true, improving the estimator’s method is like tightening the tolerances on a component that was never the source of the fault. You can produce a beautifully evidenced projection and watch it inflate anyway during the approval process, because the pressure to inflate does not come from the estimator’s ignorance. It comes from the purpose the number is being asked to serve.
The optimism-bias adjustments that have entered public-sector guidance in recent years are a genuine step forward, precisely because they acknowledge this — they treat over-optimism as a systematic tendency to be corrected structurally, not a personal failing to be scolded away. But even they are applied to the estimate, and the deeper distortion sits in the decision process the estimate feeds.
Reading the Politics Honestly
If the number is political, then the practitioner’s job is not only to estimate but to understand the politics — and that is uncomfortable territory, because it means treating the business case as an act of persuasion rather than a purely technical artefact. I would rather that were not so. But refusing to see it does not make it less true; it only makes one easier to fool.
A few things follow from taking the politics seriously.
- Separate the decision to invest from the promise of return. Much inflation happens because a single number is asked to do two jobs at once: justify the decision and set the target. When the two are fused, the pressure to justify contaminates the target. Prising them apart — deciding to proceed on strategic grounds, then setting a benefits target that nobody had to exaggerate to win approval — removes much of the incentive to lie.
- Put the estimator’s name on the number, and keep it there. Inflation thrives on anonymity and impermanence. A figure that stays attached to its author, and that the author knows will be revisited, is estimated more soberly than one that will vanish into the approved case.
- Judge the portfolio on realisation, not approval. As long as the organisation celebrates what it has approved rather than what it has realised, it is rewarding the wrong act. Shift the applause to realisation and the incentive to inflate begins, slowly, to lose its grip.
- Treat a suspiciously attractive case as a risk, not a prize. The most seductive business case in the room is often the one that has been optimised hardest against the approval criteria. Experienced boards learn to be most sceptical precisely where the numbers are most flattering.
The Honest Conclusion
None of this is a call for cynicism. I am not arguing that business cases are worthless, or that everyone producing them is gaming the system in bad faith. I am arguing for something more modest and, I think, more useful: that we stop pretending the approval number is a forecast, and start treating it as what it plainly is — a figure produced under pressure, for a purpose, by people responding to the incentives we ourselves have set.
The organisations that get closer to the truth are not the ones with the most sophisticated estimation models. They are the ones honest enough to admit that the number which secures the funding and the number which turns out to be true are, structurally, produced by two different processes — and disciplined enough to care about the second one long after the first has done its work. Until we design our approvals around that honesty, we will keep approving the most optimistic numbers in the room, and keep being surprised, year after year, that the returns never quite arrive.