Why capital projects deliver 56% below forecast

Hi Everyone,

In 2018, Lidl abandoned an inventory system built with SAP, wasting seven years and roughly €500 million. The system was never fully operational. Long before its cancellation, the project had grown too expensive to finish and too political to scrap. Failed capital projects derail years before the write-off occurs.

A project can project a strong financial return and still be a poor use of capital. These missteps usually stem from four essential questions that go unasked before financial modeling begins.

Before approving any major capital request, answer the questions below first.

Does it support a chosen strategy?

A proposal can look profitable on its own yet fund a business unit you should be winding down. Michael Mauboussin's work on capital allocation emphasizes a core principle: fund strategies, not projects. Your job is to back the few strategies worth pursuing, then fund the specific projects that serve them.

Before evaluating financial returns, align the request with an overarching strategy.

Would you still fund this project if you had to defend it as part of a broader strategic direction, rather than a standalone purchase?

If the honest answer is no, the ROI calculation is answering the wrong question.

Does it work if the results are 30% worse?

Most capital plans are built on a base case that assumes everything goes perfectly. It rarely does. McKinsey and the University of Oxford studied over 5,400 large IT projects and found they ran 45% over budget on average, while delivering 56% less value than promised.

Before building a detailed financial model, run three rough stress-test scenarios:

• Costs run 30% over budget.
• Benefits fall 30% short.
• The timeline takes six months longer than projected.

If the request clears your hurdle rate with room to spare in all three scenarios, move it forward. If it only works in the best-case scenario, re-scope the project or reject it.

Can you stop it if it goes wrong?

Execution capacity and reversibility rarely appear in a business case, yet both dictate whether you can halt a failing project. Lidl's inventory system became impossible to cancel long before its official failure because the program ran for years without clear off-ramps.

Stage-gate funding fixes this issue. At each milestone, review the evidence and decide whether to continue, pause, stop, or change direction before releasing more funding.

Whenever possible, structure the expenditure as a lease, a subscription, or a managed service so that walking away remains a viable option.

Is this the best use of the money?

Capital spent on one project is capital unavailable for another. A project worth doing in isolation can still be the wrong choice if a superior alternative exists. Review processes that only offer an "approve or reject" choice fail to surface better options.

The fix requires one extra rule: every proposal must name its next-best alternative and explain why the primary choice wins.

This requirement forces a true comparison. The next-best option might be a smaller version of the project, a delay to gather more data, or doing nothing to preserve cash. Often, the alternative proves to be the better call—but you will only discover it if you force someone to write it down.

Try this today

Take the next big spend you're considering and run it through the four checks before you decide: does it support a goal you've chosen, does it still work if things are 30% worse, can you stop it if it goes wrong, and is it the best use of the money?

Grab the CapEx Review Template we created here, fill it in, and if the capital expense only holds up when everything goes to plan, send it back to be reworked before it goes any further.

Go deeper

👉 McKinsey: Delivering large-scale IT projects on time, on budget, and on value – the 5,400-project study behind the 45% overrun and 56% shortfall, plus the four factors that separate the projects that work.

👉 BCG: The Art of Capital Allocation – the portfolio view of capital, including why growth and harvesting businesses should get very different investment rules.

👉 Bank of England: High hurdles – survey evidence that firms set hurdle rates well above their cost of capital, and why that spread sticks.

👉 Morgan Stanley: Capital Allocation – the full "fund strategies, not projects" framework and a guide to judging how a management team allocates capital.

Coming up tomorrow

Tomorrow we'll talk about which decisions you can make quickly, and which ones need more time.

That's it for today. Thanks for reading!

P.S. What's the most expensive thing your company kept funding long after it should have stopped?