Often, new projects start with a team that just has been brought together, and it’s a constant struggle for managers to make realistic commitments for new teams. When the team has never worked together before, using historical data to make probabilistic forecasts is flawed. Trying to use the data from similar teams is irrelevant, simply due to the fact that no team is the same.

Still, using Monte Carlo simulations to make probabilistic forecasts is amongst the most reliable approaches that you can take for making future predictions. Another big perk is that the simulation takes just a couple of minutes to come up with accurate results, which makes it one of the most preferred methods for setting and managing realistic goals.

So, how can we make commitments for new teams using probabilistic forecasting when we’re just starting out?

How to Make Commitments for New Teams using Probabilistic Forecasting

Any team, even new teams, can use probabilistic forecasting. This may feel like an advanced concept but it really isn’t. It’s just very different from what we’re used to.

If your whole setup is new (your team, your project, maybe even your company) and you don’t have any historical data just yet, try to buy some time from your customers before you make your final commitment. Ask for a few iterations before you provide the delivery forecast so that you can already start working on your project and collect data. Focus on managing the flow of work effectively. The more predictable your system becomes, the more accurate forecasts it produces.

If your customers are unwilling to wait, you will have to use an alternative method to provide an initial forecast (such as Planning Poker, for example). By all means, don’t make any long-term commitments for your new team at this point. You don’t want to jump into deep waters when you haven’t yet learned how to swim. The goal would be to switch to using the Monte Carlo simulation once you do have some data available.

If you and your new team are maintaining a stable system, you don’t need more than 20 to 30 completed items to produce a reliable probabilistic forecast. If you do have the data, use the data. If you don’t have the data, collect the data and start using the data.

Use Continuous Forecasting to Improve Your Chances of Hitting Targets

It’s a fact that we don’t deal well with probabilities. The fear of uncertainty is a cultural problem, and it has been proven by Professor Geert Hofstede, who conducted one of the most comprehensive studies when it comes to exploring cultural values. The trends of the uncertainty avoidance index, he asserts, are tremendous.

Geert Hofstede - the uncertainty avoidance index - commitments for new teams

More often than not, higher management takes the date and ignores the probability. Placing more value on the latter requires a shift in the mindset: we need to start thinking probabilistically and not deterministically. We need to acknowledge that forecasting is an activity in which more than one result is possible. Forecasting is all about embracing the uncertain and unknown, rather than assuming that everything will go as planned.

So, how do we deal with that uncertainty? Well, we take a probabilistic forecast as the initial answer, but we need to repeat it as we know more, as we discover more information. Remember, it’s a living forecast and it can change.

Frequently revisit and update your short and long-term forecasts, regardless of whether the number of work items in your backlog changes or not. Your performance will vary. There are plenty of factors that may affect your historical data (such as bottlenecks, blockers, dependencies and defects, just to name a few); thus, the throughput your initial delivery prediction used will change when these impediments crop up. As a result, the commitments and the probabilities that the forecast produces will adjust considerably over time.

Repetition is the key to success here. And we know that, sometimes, it can be challenging to educate your clients or stakeholders on the fact that the delivery dates can change. Having a regular cadence where you sit together and revise the forecast will probably open many doors for you. The tradeoffs on scope and time have now become data-driven decisions that are made with an explicit confidence level.

You should continually update your forecast to minimize any unexpected behavior and then adjust your course accordingly. We have a couple of strategies that we would suggest using to get back on schedule if you go off track.

Determine the Best Course of Action to Stay On Track

If your clients or stakeholders don’t like the numbers that the forecasts produce, there are a few approaches that you can use to make sure you stay on track.

  • Reducing the scope. You can commit to a project’s scope, time, or quality – but not to all three. Pick any two. If the number of items decreases, the delivery date will decrease as well. Focus on discovering the most feasible options that will still bring value to your customers. Drive your first deliverable to a true minimum viable product (MVP) and subsequent deliverables to minimum units of incremental improvement.
  • Improving your performance. And by that, we don’t mean pushing teams to work harder. On the contrary, you need to establish a stable delivery workflow optimized for predictability and reduce the waiting time in your system to a minimum. That’s the easiest and cheapest way to improve your delivery speed. In our Sustainable Predictability program, we explore the proven strategies to eliminate the obstacles that hinder your performance and enable you to deliver on time, every time!
  • Adding more capacity. Eventually, you can allocate more people to the team. By all means, there must be a clear, sensible intention justifying you moving people to different teams. To make sure you make the best out of your historical data, hold the same team together. If that’s not an option, strive to keep the working practices, skills, and expertise of the individuals fairly similar, in order to preserve your system’s predictability.

Commitments Are Not Contracts, Commitments Are Goals

Making probabilistic forecasts using Monte Carlo is simply a matter of counting the work items and running the simulation to get a range of commitments and the probabilities that come with each of them. This approach is significantly more solid and accurate than estimating since it uses your actual performance data to make a delivery prediction. 

Once you start collecting your data and putting it in action, strive to maintain a stable system to produce accurate probabilistic forecasts.

Moreover, continuous forecasting will enable you to move away from just working through the (fixed) plan. The fact of the matter is that ground conditions constantly change, and you should be able to react quickly and adjust accordingly. At some point, the direction has to change based on the new information you get.

Commitments are not contracts. These are goals. That is the first thing that you should emphasize with anyone who is asking for delivery dates.

No model provides 100% certainty that something will happen in the future. Getting rid of uncertainty and being in control of everything is impossible. By adopting a probabilistic forecasting approach, you will be able to handle the uncertain and unknown in the most effective way.

Do you find this article valuable?
Rating: 4.8 stars (10 readers voted)