Planning for half your projects to finish late sounds like giving up. But one company I worked with did exactly that on purpose, and its reasons are hard to dismiss.
When 10 of your 20 projects have blown past their target in-service dates, it’s tempting to call the portfolio underperforming. For years, I’d have agreed. Especially if those overruns fell outside the 10% tolerance window that most organizations allow.
Then a client changed my mind.
I was roughly two-thirds of the way through helping them build out schedule and financial reporting when the company’s guidance on project performance suddenly shifted.
The old guidance said no project should finish more than 10% over its approved schedule and budget (usually an aggressive bar). The new guidance said that across the portfolio, half of the projects should finish over the baseline, and half should finish under.
If the portfolio didn’t split that way, the company argued, the original estimates were suspect.
What is a p50 estimate?
A “P50” is the date or budget a project has a 50/50 chance of beating or missing. It’s basically a coin flip, and some companies (especially in engineering and construction) run careful simulations to work out what that number should be.
But where some companies would tack on 10% and move on, this company watched closely to see whether its P50s were accurate in the first place. By its logic, if the portfolio showed a bunch of projects finishing at or below that number, the P50s were probably BS (not to put too fine a point on it).
The company wanted to get the most out of its available resources and cash, and padded estimates locked both up where they weren’t needed. It wanted to be predictable more than it wanted to be early.
That guidance made one thing clear to me for the first time. Every organization has a philosophy of late, whether it intends to or not.
What is a philosophy of late?
I define a “philosophy of late” as an organization’s working answer to four questions:
- How are projects estimated?
- How are targets and contingency derived?
- How often should projects miss their targets, and by how much?
- What does a miss say about the way the project was run, and about who ran it?
You’d expect most organizations to spell out these answers. In practice, few do. So project managers are left to infer them from what gets rewarded and what gets penalized.
Every organization has a “philosophy of late.” Few have said theirs out loud.
Here’s the philosophy I run into most often. It’s probably the one you run into, too.
The team estimates the work. Leaders push back. Everyone negotiates a new estimate, and the organization adds 10% contingency to cushion whatever goes wrong. From there, most or all projects are expected to come in at or under the estimate plus contingency.
The approach is simple, and it’s built to keep overruns in check. I don’t think it’s wrong, per se. But it does reward some bad behavior.
If anything beyond 10% counts against you, the safest thing a project manager can do is pad the estimate before it becomes the baseline. Once work is underway, the safest forecast is whichever one says “on target.”
In organizations that work this way, three numbers that should stay distinct end up merging into one in people’s heads:
- The estimate. The team’s prediction of how long the work will take. Ideally, it’s a range of outcomes rather than a single number.
- The target (baseline). The date the organization manages to, often with contingency or a tolerance window built in.
- The forecast. The revised prediction once work is underway.
If you throw out the estimate once the baseline is set, the baseline becomes the only number that matters. All the context is gone. You can come in squarely in the middle of the estimate’s bell curve and still miss the target. Your performance gets dinged anyway. The result is padded estimates at the front end and Pollyanna forecasts the whole way through.
What happens when you plan for half your projects to finish late?
My client’s new guidance shifted those rewards, for better and for worse.
On the plus side, it rewarded being honest about what the numbers mean. I’ve watched companies fund projects at a P50 or P60 number (one with a 60% chance of holding), then lose sight of the risks that could push results either way. They’d tack on 10% contingency and hold the PM to that number as if it were a sure thing.
My client’s approach changed what got rewarded. In the new world, a project manager gets credit for an honest forecast. An early finish draws as many questions as a late one. Padding stops paying off, at least in theory.
Whether they mean to or not, people bend toward whatever the rules reward.
But the guidance missed something important, too. At least at first, it never said by how much a project should come in under or over. Going by the guidance alone, a project 10% over its baseline and a project 50% over were the same thing.
In a 2022 study of 4,677 IT projects, Bent Flyvbjerg and his co-authors found about as many came in under budget as over. But the average project cost 1.8 times its budget, because about one in ten cost more than double. Schedules followed the same pattern, though less extreme: most projects missed their dates by a little, and a few missed by a lot.
That means a probability curve can, on its own, offer false comfort.
Letting the forecast drift from the target is truer to what the estimate meant in the first place. But people don’t always handle nuance well, and talking in ranges can also take away the urgency to stay close to the target, or to finish early when a project can.
You need that urgency, because the bigger the portfolio, the more likely it holds a few significant outliers. The time and money saved on the many projects that stay tight can help cover the few that don’t.
Is there a perfect way to manage project deadlines?
From my own experience, I’m not sure there’s a “perfect” philosophy of late. Each one optimizes for something different. And even if you pick a philosophy to control overruns, or to make forecasts predictable, or to keep every promise, the wildcard always seems to be humans.
Whether they mean to or not, people bend toward whatever the rules reward.
No philosophy gets complicated human behavior exactly right. That’s also why no single philosophy will work at every company, or in every company culture.
The great irony is that while a bunch of us are busy arguing about numbers and probabilities, how we lead people may have the biggest effect on performance of all.
Exceptional leaders stand in the gap between what executives want and the behaviors that produce it.
How should PMO leaders counter their philosophy of late?
If you lead a practice, it’s critical to lead with your philosophy of late in mind. Every philosophy of late rewards one thing and neglects another. Your job is to reward what it neglects.
When your philosophy rewards forecasts that are always green, you can publicly recognize the PMs whose forecasts have a track record of being truly useful.
When your philosophy rewards padded estimates, you can speak up when a project is performing suspiciously well. What’s behind it? Which specific risks haven’t materialized? How realistic was the original estimate?
That’s what leadership looks like, and it’s not a “soft” skill at all.
SPIs, CPIs, S-curves, and every other KPI are only as useful as your team’s willingness to be honest. Exceptional leaders understand this. They stand in the gap between what executives say they want and the behaviors that will actually produce it.
If You Only Do One Thing
Name your organization’s philosophy of late, spoken or unspoken. Decide how you’ll counter-optimize for whatever that philosophy glosses over. If it rewards padding, ask publicly why your last early project finished early. If it rewards simplistic forecasts, ask at every status update which factors could still move the date. Teach your team to think holistically, because no philosophy will teach that on its own.