If only the number
changes, stop.

At the beginning there is a requirement, a concept, some preliminary engineering and an estimate.

Then people start discovering things.

Interfaces turn out to be more complicated than expected. Operational requirements emerge. Regulators become involved. Ground investigations reveal something inconvenient. Resilience requirements develop. Suppliers expose assumptions that nobody realised were assumptions.

And sometimes the required capability itself grows.

Havant Thicket illustrates several of these effects. The original reservoir was designed around supplying 21 million litres a day. Subsequent integration with Southern Water's wider Hampshire water-transfer plans required significant changes, including additional pipeline and interface works.

Construction also revealed ground conditions that Ofwat's technical advisers concluded could not reasonably have been foreseen from the original investigations.

So the project changed.

Quite the opposite.

Cost pressure is healthy.

Engineers can be conservative. We can prefer technically elegant solutions. We sometimes protect ourselves against uncertainty. Given an unlimited budget, there is always the danger of producing a magnificent solution to a problem that could have been solved more simply.

Management therefore has every right to challenge an estimate.

Why does it need to cost this much?

Are we gold-plating it?

Can we simplify this?

Do we really need that requirement?

Those questions can save enormous amounts of money.

The difficulty is knowing when to stop asking them.

This is one of the most dangerous moments in a major project.

Engineers do not have unlimited appetite for fighting the same battle.

If senior management repeatedly says an estimate is too conservative, contains too much contingency or represents a gold-plated solution, eventually the engineering team may stop pushing back.

Fine. We'll do our best to deliver it for £600 million.

The meeting becomes considerably easier.

The project becomes affordable. The business case can proceed.

Management may even believe that persistent challenge has finally exposed excessive engineering conservatism.

But did the engineers change their minds, or did they simply stop arguing?

The spreadsheet cannot tell you.

There is a simple discipline senior management can apply.

Whenever the estimate falls materially, ask: What changed?

If a requirement disappeared, the design became simpler, a survey provided better information, a risk was genuinely retired, a supplier demonstrated a cheaper method or a prototype proved an assumption, then the project really may have become cheaper.

Keep challenging.

But eventually the answers start sounding different.

We have reduced the contingency.

We think we can manage the risk.

We are assuming that won't happen.

The contractor has accepted it.

We'll resolve that later.

Those statements may be perfectly legitimate project decisions.

But they are not necessarily cost reductions.

Sometimes they merely move uncertainty somewhere else.

And moving a risk does not make it disappear.

When a project originally associated with a few hundred million pounds eventually approaches twice that amount, newspapers understandably describe it as an overrun.

Sometimes that is exactly what it is.

Poor estimating exists. Weak project control exists. Gold-plating exists. Bad procurement exists. Uncontrolled scope creep certainly exists.

But there is another possibility worth examining.

Some of the eventual cost may always have been inherent in delivering the capability.

It simply wasn't visible, understood or acceptable when the first number was produced.

When did we actually know enough about this project to know what it was going to cost?

If the answer is after we had already committed ourselves to building it, the problem started long before the cost overrun appeared.