Why Projects Fail: Three Checks Before the First Invoice

Less than half of all projects actually stay on budget.

You have seen it from the inside. The status slide is green every Monday. Then one Thursday the money is gone, and the deadline moves for the third time.

Here is the irony. The team logs every hour to the quarter. Nobody wrote down, before the start, where "a bit over" ends and "stop" begins.

The villain is not the team. It is hope: the plan that never checked its own starting point.

A green slide that turns red overnight belongs to a plan that never checked where it started. The Jirgenson Model checks the goal, the starting numbers and the environment first, then sets red lines for budget and deadline. It is scientifically proven that the Jirgenson Model works (doi.org/10.5281/zenodo.23964934). You know where stop begins before the money goes.

Project over budget: draw the red line before day one

Fewer than one project in ten lands on time and on budget.

The first number says budgets slip. This one says budget and deadline rarely hold together. Both point to the same gap: nobody set the limit in advance.

Fix it on day one. Write the budget and the deadline as goals. Give each one six fields: parameter, definition and source, target and deadline, current value, permissible deviation, owner and check rhythm.

Then give each one three zones. Green: within tolerance, leave it alone. Yellow: worse than normal, find the cause while it is cheap. Red: past the limit, everything else waits.

Andrew, who runs a coffee shop called "The Corner", set his red line for the cash reserve at one month. One month gives him time to negotiate, find money and decide. Below that, there is no time for any of it.

Add one leading indicator: a figure that moves before the result. A budget report tells you what already burned. A leading indicator rings before the fire.

How to avoid it: set the zones in a calm hour and do not move them mid-crisis. Moving the line in a storm bends the yardstick to fit the result.

Why projects fail before the first task: the three-offset check

One project in two hundred delivers on budget, on time and with what it promised. One.

Budget, time and scope together, in full, is the rarest outcome there is. So decide now which of the three gives way first. That is the refusal test. If you had to give up your goals one by one, which would go first? Which would be the last one left?

Then run the check that catches the deepest errors. Every plan has three anchor points: the goal, the starting point and the environment. Shift any one of them, and no amount of effort saves the plan.

Here is the full check. Write one answer per row.

Now the same check, filled in. In month seven, with business rising, Andrew ran it on his biggest project: a second location.

For offset (c), ask four more questions before you call an environment doomed. Has the decline lasted several periods in a row? Are the strongest players leaving? Do the numbers add up in at least one niche within your horizon? What would it cost to leave by mistake?

That same month, an agent offered Andrew a renovated unit in a shopping centre at an attractive price. Andrew asked about the ship, not the deckchair. Footfall had fallen for three years. The anchor tenant had left. He said no.

Any row where you write "I don't know" is a risk you have found. The whole check takes fifteen minutes.

Cost overrun starts with a wrong starting number

Big IT projects run almost half over budget and deliver less than half the value they promised.

What is cost overrun? Spending more than the budget the project started with. The budget sat on the starting numbers. When the starting numbers are wrong, the overrun was written in on the first day.

Audit the start before you plan. Compare every number in the plan with the data in the systems, not with your impression.

The coffee shop shows the scale. Andrew planned to hire a second barista for €1,800 a month. At the margin he believed, twenty per cent, the hire left €4,200 of profit. At the real nine per cent, it left €900. The road to a two-month reserve stretched to almost fifty months.

An eleven-point error in one input turned a sensible decision into a four-year hole.

How to avoid it: no new spending until the starting number is the measured one. Call it the planning fallacy or plain hope. Either way, the plan starts on a floor you are not standing on.

Project risk that can sink the whole company

One IT project in six goes so wrong it threatens the entire company.

A project should never be able to sink the business that pays for it. So write the business's survival above the project, in the goal list and in the constraints.

Constraints come first. A constraint answers one question: what will I not do, even if it would save the goal? Andrew wrote two: never cheaper beans, never late wages. Write yours for the project. "The project never takes the cash reserve below one month" is a good start.

When a number crosses its red line, follow the red protocol. Check for any direct threat to people, payments or the law. Freeze secondary work. Put time, money and attention on the red number. If several numbers turn red, save first the one that threatens the most imminent irreversible damage.

How to avoid it: the project row never outranks the cash row. Your order says so before the crisis, so fatigue cannot say otherwise later.

Project delays: what six more months of hoping cost

Delay has a price, and it has numbers. Take Andrew's reserve. At a 9% margin it grows about 0.1 months a month. If nothing changes, the reserve reaches its goal of two months at month 16, against a deadline of 12. At a 15% margin it grows about 0.18 a month, and the goal arrives at month 10.

Six extra months, one breakdown away from trouble. At 9%, a breakdown worth 0.1 months of reserve pushes the goal back about a month. At 15%, about half a month.

Now picture the second location signed on impulse in month six. The new rent eats a reserve that has not yet reached two months. It falls back to red. Six months of work, gone in one.

And the human cost: another quarter of status meetings where everyone knows, and nobody says it.

"My project is far bigger than a coffee shop"

The scale changes. The anchor points do not. A medium-sized company books €100,000 of stock in its warehouse. A physical stocktake finds half of it out of date or spoiled. The real state is €50,000. Same offset (b), with more zeros.

The same company plans sales up 50% with its credit lines fully drawn and no money for raw materials. That is offset (a): one of the two goals has to go. The check works for a workshop, a software rollout or a family budget.

Where the three checks stop

Be clear about what the check gives you. It treats the symptom: you stop starting projects that were broken on day one.

The deeper cause is still in place. A plan checked at the start can still drift in month four. You do not yet know how far it may stray at each checkpoint, or when to stop it. And the plan runs forward from today, not backward from the finish.

Two steps close that gap.

• The tolerance corridor and launch conditions. How far you may stray at each checkpoint, and the written condition for starting, stopping or revising. Chapter 10.
• The backward move. The route from the finish line to this week, with resources checked at every step. Chapter 15.

The three checks sit at the start of one loop: goal, state, error, levers, plan, forecast. Without the next two steps, the check tells you the plan can work. It does not yet keep it working. For the same check on a business plan, read why business plans fail.

Two choices for the next kickoff

You have two choices. Start the next project the usual way, trust the green slide, and learn about the overrun on a Thursday. Or spend fifteen minutes on three questions before the first invoice.

The check tells you the plan can start. The corridor, the launch conditions and the backward move keep it on course after day one.

Next Monday you open the project page and see one line in red, three months before anyone else would.

Test the logic for free: Chapter 1, with the refusal test. Or get the full control system on Kindle, on Amazon. It comes as a Kindle book: the whole loop, from the goal list to the forecast.

Published research

It is scientifically proven that the Jirgenson Model works. The published experiment, its record and its data:

• Zenodo (DOI): 10.5281/zenodo.23964934
• Hugging Face: E5 Jirgenson Model blind test
• Kaggle: E5 blind test dataset
• GitHub: experiment folder
• Wikidata: dataset Q142022866, author Q141981336
• ORCID: 0009-0001-3034-3552