Why You Need Probabilistic Forecasting as a Project Manager
Hey there, my name is Sonya Siderova, founder and CEO of Nave, and my main goal is to make your job as a project manager more efficient and less stressful. Here at Nave, we do that by sharing solutions to help your teams work smarter (not harder), so everyone can go home happy.
If you’re struggling to deliver on your commitments, it may be time to break the status quo and try a new solution. We’ve found a lot of success with probabilistic forecasting and today we’ll explore how it works.
“Trust your gut,” is usually good advice.
Except when it isn’t.
Trusting your gut makes sense when it comes to choosing which size T-shirt to order for your best friend’s birthday gift.
Trusting your gut is not so great when it comes to project forecasting.
That’s because (surprise) we humans aren’t perfect, despite our best efforts. We miss deadlines, not because we aren’t trying hard enough, but because it’s actually quite challenging to gauge how long a project will truly take.
What’s one of the most reliable solutions, then, to make sure your deadlines are realistic?
Now, you might be wondering: “Wait, ‘probabilistic’ sounds like there’s a degree of uncertainty involved”, and uncertainty is definitely something most people aren’t comfortable with. How is that any better than relying on your gut or your best guess?
Let me take a step back and start by sharing with you my super simple secret to a happy marriage (that’s lasted over 10 years now): regular date nights!
My husband and I have date night every single Friday – we’ve found that this little ritual is what helps us stay connected with each other, feel close, and escape from the hustle and bustle of the week.
If you’ve been around for a while you probably already know that I’m a huge fan of Michelin-starred restaurants. As such, most of our dates take place in our favorite place “Bar Vaccin”.
On our most recent date, we took a seat as usual at the bar and enjoyed a couple of drinks while our table was being prepared. This time, the staff gave us something new while we waited: a device that told us how many minutes were left until our table was ready.
Honestly, we didn’t expect it to predict exactly when we would be seated, but it ended up being quite accurate. In fact, the table ended up being ready even sooner than we were told (a good thing, considering how hungry and excited I was to taste those first appetizers!).
Of course, my first response was to talk to the owners of the restaurant to figure out what’s behind that small piece of electronics.
It turned out that the device worked so well because it relied on past data – a range of wait times – in order to give us a realistic estimate of when our table would be ready for us.
Imagine my excitement! And, yes, I’m biased, but I’ll be saying this over and over again. You have to use your past performance data to come up with reliable forecasts, especially when it comes to project management. There is just no other way around it.
Michelin-starred restaurants are world famous for their impeccable quality and attention to detail. I loved that new initiative and I was really impressed by the approach they’ve taken to bring that place to a whole new level.
When it comes to project forecasting, you want to use that same principle. Let’s take a closer look at how it works.
What is Probabilistic Forecasting in Project Management?
Because it’s impossible to predict anything with 100% certainty, the next-best option is to use your past performance data to come up with a range of project delivery dates and the probability of hitting each of them.
Here is what probabilistic forecasting isn’t: “We’ll deliver these 50 items by November 1st!”.
What’s the problem with this statement, you might ask? There are two major issues with it.
Firstly, it suggests that there is an actual obligation to deliver all of the 50 items by November 1st. And secondly, it implies that these exact 50 items will be finished by then.
Can we make that promise knowing that knowledge work is notorious for its unpredictable nature?
We should always remember that there is uncertainty involved and we should communicate that there is more than one possible scenario that might happen in the future.
So, a probabilistic forecast would look like this: “We can deliver 50 items by November 1st with an 85% probability of hitting that target.”
The difference? You can take any 50 items from your backlog (and yes, feel free to add and remove from that list!).
Furthermore, you communicate the risk you are managing in terms of percentages. There is an 85% chance you will make it on time.
This is what probabilistic forecasting is – and it’s a method that’s actually far more effective, accurate and reliable than just about any other.
Probabilistic forecasting doesn’t provide perfect predictions.
It provides reliable predictions. Remember, there’s no such thing as a perfect prediction – alas, no one has invented the magic crystal ball just yet, unfortunately.
You can only manage the uncertain and unknown effectively and mitigate the risk of failure. And there is a tool at your disposal that will enable you to do just that.
How to Make Reliable Probabilistic Forecasts?
The tool to make reliable probabilistic forecasts for your project (in just a couple of minutes!) is called the Monte Carlo simulation.
This simulation uses a computational algorithm to provide you with a range of possible delivery dates and the likelihood of reaching each of those delivery dates.
You just enter when you want to start your project and the number of items in your scope and the simulation will build a histogram with percentiles increasing from left to right.
Let’s say the scope of your project is 50 tasks and you want to start working on it on September 13th. The simulation will give you a range of possible dates for completing your project.
In this example, there is an 85% likelihood of finishing your project by February 28th, compared to a 95% likelihood of finishing your project by March 13th.
Remember, whether you are using probabilistic forecasting or a different approach, there is no 100% guarantee of completing your project. But Monte Carlo, unlike other approaches, will give you the most probable, least-risky outcome to commit to.
You can choose to take a higher risk to reach an earlier deadline or mitigate your risk by choosing a more realistic deadline. Ultimately, you are in control of how much risk you and your team feel comfortable taking and are empowered to choose from your options.
“When will you deliver your project?” may not be the right question to ask. Monte Carlo will give you that answer. The question now is, “How much risk are you willing to take?”.
Probabilistic forecasting can be a counterintuitive approach to adopt at first because people are naturally averse to the notion of “probability.” If it’s not a sure thing, why are we relying on it?
The truth is, Monte Carlo allows us to get as close as possible to a “sure thing” by running thousands, even millions of trials to get a range of tasks and the confidence levels that come with each of them.
There is just one prerequisite here: to make accurate and reliable predictions with Monte Carlo (and actually make any other approach to forecasting work), you need to optimize your delivery system for predictability.
Why is Forecasting Important in Project Management?
Do any of the following sound familiar:
- Despite your best efforts, you and your team continuously underestimate how long certain deliverables will take
- Because things take longer than expected, it becomes harder to trust that you’ll ever actually meet the deadline
- As a result of the above, you and your team members feel frustrated and worried that you’re losing credibility. Meanwhile, your clients and stakeholders are also concerned that things aren’t getting done…
Probabilistic forecasting changes the game because it’s fast, cheap and most importantly reliable.
The more predictable your delivery system is, the more reliable probabilistic forecasts it will produce.
You’re effectively taking out the guesswork and you’re managing risks effectively. You are empowered to choose which outcome makes the most sense for you, based on the % of risk you’re willing to live with.
If you’re willing to explore the proven roadmap to optimize your delivery workflow for predictability, I’d be thrilled to welcome you to our Sustainable Predictability program.
Probabilistic vs Deterministic Forecasting
Deterministic forecasting is just that: determined. Fixed.
A deterministic model comes up with a single date for your project completion. A deterministic forecast limits you and is more likely to fail for one simple reason – the future is not deterministic. At least, thinking about it deterministically is certainly not enough.
Traditional deterministic thinking has instilled deep roots in our mindsets. We have been committing to single certain delivery dates for so long. Being uncertain feels very uncomfortable.
The biggest problem with deterministic forecasting is that we assume we know what will happen in the future with 100% certainty upfront. 100% certainty for the future does not exist in our world.
Adopting the concept of probabilistic forecasting, generally speaking, means accepting that there is more than one possible result that may happen in the future. To provide a reliable delivery commitment, you need to communicate a range of delivery dates and the probabilities that come with each of them.
Remember, the main purpose of making estimations is to manage risks effectively. There’s always risk involved and the most effective approach to managing realistic expectations is to quantify that risk by assessing the probability of achieving your goals.
Every delivery prediction should communicate the risk associated with it and probabilistic forecasting is one the most reliable solutions.
How to Get Started with Probabilistic Forecasting
At this point you may be thinking: “I get why probabilistic forecasting makes the most sense for my project – but what if I don’t have enough data right now?”
Then you go and get the data. You don’t need a year’s worth; quality over quantity is what matters here. 20 to 30 completed items should be enough to get you started.
Keep in mind that if your team is totally new to using probabilistic forecasting, you’ll want to introduce it gradually. Just like Rome wasn’t built in a day, you won’t want to change your method overnight – your team needs time to transition and adapt.
Remember, the one and only prerequisite to making probabilistic forecasting work is to optimize your system for predictability.
As you take control of your management practices, the numbers and results will come to speak for themselves and you, your teams and clients will notice your performance improving and your project delivery commitments becoming much more predictable.
Where to Go From Here…
Hopefully, at this point, I’ve convinced you that probabilistic forecasting is faster, cheaper, and more reliable than any other approach, and only takes a couple of minutes to try – so there’s really nothing to lose by checking it out and seeing if it works for you and your team.
If you’re ready to take the next step, go ahead and connect Nave to your platform: The tool will analyze your data and within just a few minutes you’ll be set with your very own probabilistic forecast for your current projects. We offer a 14-day trial for free →
Now, here is what I want you to remember from this article:
- Don’t rely on judgment, intuition, gut feeling or any other relative complexity measurement for coming up with a project deadline, do use your past performance data
- Don’t reconcile with a single, fixed delivery date – make sure you use Monto Carlo so that you have multiple outcomes you can work with
- Do understand the % risk with each deadline, and choose the risk you’re most comfortable with
- Strive to optimize your delivery system for predictability to improve the accuracy of your probabilistic forecasts
I hope you found this article helpful. If you have a friend or a colleague who’s struggling to deliver on their commitments, please share this article with them – I believe it will make a world of difference in reaching their goals.
That’s all for today, my friend! Thanks for reading (and sharing), and I look forward to seeing you here next week, same time and place, for more managerial goodness. Bye for now!
Meet the Author
Sonya Siderova is a passionate product manager and a driving force behind Nave, a Kanban analytics suite that helps teams improve their delivery speed through data-driven decision making. When she's not catering to her two little ones, you might find Sonya absorbed in a good heavyweight boxing match or behind a screen crafting a new blog post.