There is a sentence that is very hard to say in a project review, even when it is the right one.
I don’t think we should do anything.
Say it and watch the room. Somebody will assume you have not understood the problem. Somebody else will remember it in three months if anything goes wrong. Nobody has ever been criticised for mobilising a second crew, and nobody ever will be — because the cost of acting unnecessarily is invisible, while the cost of not acting is a story with your name in it.
So we act. We add resources to schedules that were already recovering, authorise overtime that changes nothing, pay to expedite deliveries that were not on the critical path. Not because anyone is foolish, but because do nothing is the only recommendation you cannot defend with a feeling.
You can defend it with arithmetic.
The asymmetry that drives the spending
Consider what happens after each decision.
You spend, and the project recovers. The intervention worked. Everyone agrees you handled it. Nobody will ever know whether it would have recovered anyway, because you removed the possibility of finding out.
You spend, and the project does not recover. At least you acted early. The problem was worse than anyone thought.
You don’t spend, and the project recovers. Nothing happens. No credit, because nothing visible occurred.
You don’t spend, and the project does not recover. This one has a meeting about it.
Three of those four outcomes protect the person who spent. Only one exposes them. Faced with that, spending is not irrational — it is what the incentives ask for. The only way to change the behaviour is to change the evidence, so that do nothing stops being a judgement call and becomes a calculation somebody else can check.
What makes it calculable
Project delay recovery is measurable: a delay that is closing is closing at a rate, and a rate can be measured. Measured against the plan rather than against last month, that rate is Earned Schedule.
Take the delay percentage on a task at each status date and compare consecutive measurements. If the gap was 26%, then 21%, then 16%, it is shrinking by roughly five points a week. That number is the whole argument.
Because once you have it, the date the delay reaches zero is division:
Remaining delay 20%, closing at 5% per week → back on schedule in four working weeks
That is not a forecast, a target or an expression of confidence. It is what the last three measurements imply if the team keeps doing what it has been doing. And it is checkable: anyone in the room can look at the same three numbers and get the same answer.
Then compare it with what action costs
Now the decision has two priced options rather than one instinct.
Option A, intervene. A second crew for a month. Weekend overtime rates. An expedited delivery with a premium on it. Whatever those come to on your project, write the figure down, and add the part nobody costs: the disruption of onboarding people into work already in progress, and the management attention consumed by running the intervention.
Option B, wait four weeks. Zero.
If your contractual milestones can absorb four weeks, option B delivers the same outcome as option A at no cost. Not a cheaper outcome — the same outcome, reached by letting a team that is already performing continue to perform.
That is the entire case, and it fits on one slide.
The honest constraint
This only works when the recovery lands inside the window you actually have.
Sometimes it does not. The calculation says four weeks and the contractual date is three weeks away, and then you intervene — but you intervene knowing something valuable: you need to buy one week, not four. You are no longer insuring against an unquantified risk; you are purchasing a specific, measured amount of time, and you can size the intervention to it.
That distinction is worth real money on its own. Most emergency spending is not wrong in direction, it is wrong in magnitude — a full crew hired for a month to solve a problem that needed one week of overtime, because nobody knew how big the gap actually was.
There is a second constraint worth stating plainly. A recovery rate calculated from two measurements is not reliable; three is the minimum before you would put your name to it, and a single unusually good week is noise, not a trend. If somebody in the meeting asks how confident you are, the honest answer depends on how many status dates the number is built from — and being able to give that answer is part of what makes the recommendation credible.
How to bring it to the meeting
Three things, and none of them takes long to prepare.
The three measurements. Not a chart, not an index — the actual delay percentage at three consecutive status dates. This is the evidence, and it is the part people can verify themselves.
The date. When the delay reaches zero at the current rate, stated as a date rather than a duration. Back on plan by 12 September lands differently from about four weeks.
The contractual window. The milestone that matters and how much room there is between it and that date. This is what turns the analysis into a decision, and it is the part that decides whether the answer is really do nothing.
If the room can see all three, the conversation stops being about confidence and starts being about arithmetic. And arithmetic is much harder to override with seniority.
What this does over time
The first time you recommend doing nothing and are proved right, it registers quietly. The third time, something changes: people start asking what the numbers say before they ask for budget.
That is worth more than the money saved on any single decision. A team that can distinguish a project that is failing from one that is recovering stops treating every deviation as an emergency — and gets to spend its attention, and its budget, on the ones that genuinely are.
The projects that need help still get it. They just get it faster, because you are not spending the same attention on the ones that were always going to be fine.
This piece assumes you already measure how your delays move. If that is new, start with Your project delay has a direction, not just a size, which explains the method and how to do it by hand.
GanttDerivata records the measurement at every status date, calculates the recovery date for every task that is closing its gap, and puts both in front of you before the review — so the case for doing nothing is already prepared when somebody asks.
#ProjectDelay #ProjectControls #MicrosoftProject #PMO