The seven signs a project is going to lose money, months before anyone notices
A project does not start losing money in the month the report shows it. It starts long before, and it leaves a trail. The problem is that the early signs are operational —things you see in the field— while the confirmation is accounting, and it arrives thirty to ninety days late. By the time the number appears, the margin is gone and the only options left are the expensive ones.
An output deviation that starts in week three can be corrected in week four. The same deviation found at quarter close can only be negotiated. That is the whole difference between watching the operational signs and waiting for the number.
These are the seven that show up first, in rough order of how much warning they give.
1 · Headcount on site drops with no explanation
Lead time: 3 to 7 days.
This is the leading indicator par excellence, and almost nobody watches it. Progress is a lagging indicator: by the time it falls, the problem happened days ago. Attendance falls today.
A crew that goes from twelve to seven people two days running almost always means the subcontractor moved people to another job, usually because they get paid better or faster there. If you find out through the progress figure, you found out late.
What to watch: headcount by work face, daily, against last week.
2 · Output separates from budget and never comes back
Lead time: 2 to 4 weeks.
Every crew starts below budgeted output: that is the learning curve and it is normal. What is not normal is that it is still flat in week three or four.
The rule of thumb: a 10 to 20 % deviation at start-up is expected. More than 25 % sustained over three weeks is structural, and projected onto the remaining quantity it already has a number attached.
What to watch: actual output against budget, by item, weekly. And above all, the trend.
3 · Lost days become routine
Lead time: 2 to 6 weeks.
One lost day is an event. Four lost days a month, every month, is a structural cost nobody is accounting for, because each one was justified separately.
There is also an attribution problem here: if you do not record the cause the same day, all those days end up being yours by default, even when half of them were not.
What to watch: lost days per month with their cause, and what share is attributable to third parties.
4 · Work faces get opened and never closed
Lead time: 3 to 6 weeks.
This is a subtle pattern and a highly predictive one. When there is pressure to show progress, new faces get opened instead of finishing the open ones. There is movement, there are people busy, and the progress percentage ticks up a little in a lot of places.
The problem is that you only get paid for what is finished, and an open face consumes haulage, scaffolding, supervision and travel without generating an invoice. It also multiplies rework.
What to watch: how many faces are open at once, and how many were closed during the month.
5 · The rejection rate on payment applications climbs
Lead time: 1 to 2 months.
When payment applications, certificates or minutes start coming back with objections, it is almost never an isolated back-office problem. It usually reflects that field measurement has degraded: measuring late, from memory, with no prior sign-off.
And the financial effect is twofold: you get paid later and you get paid less, because unsupported quantities end up being conceded to unblock the payment.
What to watch: of the last five applications, how many drew objections and of what kind.
6 · Material consumption appears with no explanation
Lead time: 1 to 2 months.
When actual consumption of a material exceeds the budgeted figure plus its waste allowance, there is a cause: poor execution forcing rework, theft, waste from bad storage, or an estimating error.
All four call for different treatment and none resolves itself. What they have in common is that you detect them by comparing actual against theoretical consumption, and that comparison almost never happens until close-out.
What to watch: cumulative consumption of the three or four heaviest materials, against what was budgeted for the progress actually built.
7 · The site engineer stops reporting in detail
Lead time: variable, and this is the uncomfortable one.
When someone who used to report specific quantities starts reporting «still progressing on the same face», there is usually a reason.
Sometimes it is workload. Sometimes the report has become a task with no return. And sometimes —it is worth saying out loud— there is something they do not want visible yet.
In all three cases, losing granularity in the report is itself a sign, because it blinds you exactly when you most need to see.
What to watch: the level of detail in the report over time, not just whether it exists.
What they have in common
All seven signs are detected from the same input: a daily record, detailed, captured in the moment.
None of them requires a sophisticated cost control system. They require someone writing down every day what was built, with how many people, on which face, what was received and what stopped. With that, all seven compute themselves.
Without it, all seven are invisible until the accounting number appears, and by then the margin is gone.
The minimum dashboard
If you had to review five numbers a week, these:
| Indicator | Frequency | Alarm |
|---|---|---|
| Headcount by work face | Daily | Sustained drop, 2+ days |
| Actual vs budgeted output | Weekly | Below 0.80 for 3 weeks |
| Lost days and cause | Weekly | More than 3 a month |
| Faces opened vs closed | Weekly | More opened than closed, 2 weeks |
| Objection rate on applications | Per application | 2 in a row |
Frequently asked questions
- Isn't it simpler to just look at the monthly cost report?
- It is simpler and it arrives late. Cost control tells you that you lost; the operational signs tell you that you are going to lose. They complement each other, and only one of the two leaves room to react.
- Which of the seven is the most predictive?
- Attendance, on lead time alone. But the most actionable is output against budget, because it comes with a number you can project onto the remaining quantity of the item.
- What do I do if I spot three or four signs at once?
- They are rarely independent. They usually share a root cause, and the most common ones are supply, work faces not released and labour turnover. It is worth hunting for the common cause rather than attacking each sign on its own.
- How do I convince management to look at this?
- With a case of your own. Take a closed project that lost money, reconstruct when the signs appeared and compare that date with the date management found out. The gap is usually months, and it needs no further argument.
- Do I need a cost control system to detect them?
- No. All seven are derived from the same input: a detailed daily record, captured the same day. What you need is the habit of recording, not the cost software.
Chilean, designing for Latin America. Field research surfaces what actually matters to clients, and that becomes products non-technical people adopt on their own — legal, education, accounting — and that show up in productivity from week one.
See all their articles →
