You have a strong case. The benefits are three times the cost, the sponsor is behind it and the team is ready. Then the investment committee, the board or HM Treasury sends it back for more work, or funds a fraction of it.
It is tempting to blame bureaucracy. More often, the approvers are doing their job. They have read hundreds of business cases, and they know how the numbers usually turn out. This article sets out what they know, how they discount for it, and what the cases that get funded do differently.
Approvers have seen this before
The evidence on technology projects is consistent, and it is not flattering.
- A study by McKinsey and the University of Oxford of more than 5,400 IT projects found that large ones ran on average 45% over budget and 7% over time, and delivered 56% less value than predicted McKinsey.
- Flyvbjerg and Budzier studied 1,471 IT projects. The average cost overrun was 27%, which sounds manageable, but one project in six was a “black swan”, with an average cost overrun of 200% and a schedule overrun of almost 70%. Some took their companies down with them HBR.
- A later study of 5,392 IT projects found that overruns follow a power law: extreme outcomes are far more likely than a normal spread would suggest, and there was no significant difference between government and private sector projects JMIS.
- Across a database of more than 16,000 major projects of all kinds, Flyvbjerg and Gardner found that only 8.5% came in on budget and on time, and just 0.5% on budget, on time and with the benefits promised How Big Things Get Done.
- In the UK government’s major projects portfolio, around three in four projects were rated amber or red for delivery confidence in 2026: 109 amber and 34 red out of 189 NISTA.
An approver who has seen these numbers cannot take an optimistic plan at face value. If your case reads like every other case, it will be discounted like every other case.
The Treasury builds the discount in
HM Treasury makes the discount explicit. Its guidance on optimism bias tells appraisers to raise cost and time estimates for known bias, starting at an upper bound and working down only with evidence that the risks have been dealt with. For equipment and development projects, the category that covers software and systems, the upper bound is 200% on capital cost and 54% on duration, and the lower bound is 10% HM Treasury.
The new Green Book, published in February 2026, goes further in the right direction. It asks appraisers to base their adjustments on their own organisation’s record of past forecasts and on comparable proposals, using the generic figures only where that evidence is missing. It says a benefit-cost ratio must not be used as “a crude pass/fail test”, and that the “business as usual” option must always be carried forward as the benchmark everything else is compared against HM Treasury.
Private sector boards rarely write this down, but they do the same thing in their heads. The difference is that you do not know what discount they are applying.
What a fundable case does differently
It takes the outside view
Psychologists Daniel Kahneman and Amos Tversky showed that people forecast from the “inside view”, the details of their own plan, and ignore the “outside view”, how similar projects actually turned out. Reference class forecasting, developed by Flyvbjerg and first used in practice for the UK Department for Transport, corrects for this by starting from the outcomes of comparable projects and adjusting from there European Planning Studies. A case that shows where its estimate sits against comparable projects answers the approver’s question before it is asked.
It gives ranges, and says where it sits in them
A single number invites a single discount. A range, with the likely outcome and a cautious one (often written P50 and P80), shows you have thought about what could go wrong, and lets the approver decide how much risk to fund.
It shows how risk was reduced
The Treasury guidance rewards evidence. Fixed requirements, an off-the-shelf product, a supplier with a track record, delivery in phases: each one is a reason to reduce the uplift, and a case that shows them earns a smaller discount.
It includes honest options
If the only options are “do nothing” and “do everything”, approvers assume the case was written backwards from the answer. A credible “do minimum”, carried forward and costed properly, makes the preferred option more convincing, not less.
It names who owns each benefit
The National Audit Office’s review of what delivers value from major projects starts with being clear about the value you want, not the project, and evaluating whether you got it NAO. Benefits with a named owner and a measure are believable. Benefits that belong to “the organisation” are not.
A worked example
A £5m software project is planned to take 18 months.
| Step | Cost | Time |
|---|---|---|
| The case as first written | £5.0m | 18 months |
| Treasury upper bound (+200% cost, +54% time) | £15.0m | 28 months |
| With evidence that 60% of the risk is managed (uplift 80%) | £9.0m | 22 months |
| Comparable IT projects, on average (+27%) | £6.4m |
The strongest version of this case does not hide from those numbers. It says: our estimate is £5m; comparable projects suggest we should plan for £6.5m and fund to £7.5m; here is the evidence that we have managed the main risks, and why the uplift should be lower than the guidance’s starting point; here is the do-minimum option and what it would cost; and here is who owns each benefit. That case is harder to reject, because it has already done the approver’s job.
Before you submit
- Find your reference class. What happened on the last five projects like this, here or elsewhere?
- Show a range. Give the likely cost and a cautious one, and say which you are asking for.
- Earn your discount. List the evidence that the usual risks have been reduced.
- Cost the do-minimum properly. Make it a real option, not a straw man.
- Name a benefit owner and a measure for every benefit you count.
- Plan to stop. Agree the points at which the project will be reviewed, and what would end it.
If you are preparing a business case for a major investment and want it to survive scrutiny, talk to us. We will test it the way your approvers will.
Sources
- Bloch, M., Blumberg, S. and Laartz, J. (2012). Delivering large-scale IT projects on time, on budget, and on value. McKinsey & Company, with the University of Oxford.
- Flyvbjerg, B. and Budzier, A. (2011). Why your IT project may be riskier than you think. Harvard Business Review.
- Flyvbjerg, B. and others (2022). The empirical reality of IT project cost overruns: discovering a power-law distribution. Journal of Management Information Systems, 39(3).
- Flyvbjerg, B. and Gardner, D. (2023). How Big Things Get Done. Currency and Macmillan.
- National Infrastructure and Service Transformation Authority (2026). NISTA major projects annual report 2025-26.
- HM Treasury. Green Book supplementary guidance: optimism bias.
- HM Treasury (2026). The Green Book (2026).
- Flyvbjerg, B. (2008). Curbing optimism bias and strategic misrepresentation in planning: reference class forecasting in practice. European Planning Studies, 16(1).
- National Audit Office (2024). Lessons learned: delivering value from government investment in major projects.
Want to talk this through?
We help organisations put ideas like these into practice. Tell us what you are working on and we will help you find a practical next step.
