Get (some of) your life back from meetings using project forecasting

Published in

The Startup

4 min readOct 5, 2020

Meetings. Meetings everywhere. It feels like we have been back to back since the start of the lockdown. What if there was a better way? There is, you can at least ditch project estimation meetings. And start using forecasting instead.

In many companies you are expected to give a delivery date to a project. And normally that involves a lot of work. A lot! You estimate requirements. Check people’s availability. Identify dependencies. Do maths. And BOOM! There is your delivery date.

This is often so time consuming that people even avoid re-estimation.

Forecasting is not only faster, and cheaper. It is also painless. And often more reliable. It’s so simple that you won’t think twice in re-forecasting.

What is forecasting

In a nutshell it’s like a formula. You input some info such as historical data and number of features. And it spits out a delivery date. Not just a date. A date with a degree of confidence!

You are a bit sceptical, right?

I was sceptical at first too. Every imaginable field uses it, from forecasting stock prices to optimizing supermarket inventories and finding best driving routes. The most popular forecasting technique is Monte Carlo simulations.

How it works

It’s all about statistics and small samples. But there is nothing fancy about it. It’s 100% brute force. It uses a combination of many, many, many possible outcomes to tell you which one is more likely.

For example, say your team needs to code some 50 features and you want to forecast how many weeks it will take to get it done. It’s easy when you know the number of features the team has delivered in past weeks. What “Mr Monte Carlo” will do is roll the proverbial dice and pick a number from past weeks. Then roll it again. Then again. And again, (…) Until you reach 50 features. Let’s illustrate it this way:

Number of features done in past weeks: 8, 7, 12, 15, 14, 7, 10 

1st roll: 12 (picked at random)
2nd roll:  7 (at random)
3rd roll:  8 (...)
4th roll:  7
5th roll: 12
6th roll: 10

Total: 12 + 7 + 8 + 7 + 12 + 10 = 56

You now know that it will take 6 weeks to complete the 50 features. Sweet! But that’s only one run. Now you run this rudimentary algorithm many many times. And that will give you many many possible delivery dates. Being 85% confident of a delivery date sounds pretty sweet, right? So lastly you need to find the 85th percentile. That is the number in which 85% of your results fall below it. That’s your delivery date.

Of course you won’t be doing this manually. You could. But you are not crazy. Troy Magennis has a few amazing spreadsheets that will do all the hard work for you. They even give you some sophisticated ways to fine tune the simulation. Use this file if you have already started the project. Or this if you have not.

When to use forecasting

Short answer: when Big Boss asks “when will your team deliver the project?”

You can do it at the beginning of the project. Or at any point when you need a re-forecast. And it helps if you can at least write up the feature titles for the remaining of the project.

When NOT to use it

Short answer: team planning meetings.

Why? Team dynamics. Most teams will benefit from using an estimation technique like planning poker. It helps create a shared understanding of the work. It fleshes out better requirements. It helps people jell. So, do the estimations. But ditch the estimates. Get it?

How to claim (some) of your life back

Use forecasting! But you will need to convince your team (and Big Boss) of the value of forecasting. How do you do that?

My preferred approach is going rogue first. Don’t tell anyone. You want to convince people with data. So you want to show them that forecasting is either a) better, or b) as reliable as estimation. A great way to do it is by letting people do the estimation their usual way. And you doing forecasting by yourself. Note both the results. Then after a while compare where you are, and where the estimation said you would be. Ditto with forecasting.

Now tell everyone!

This will get you a long way in convincing your colleagues. So, start playing with forecasting. And maybe. Just maybe. You will get some of your life back from those endless meetings.

You might also find useful:

Want to connect? LinkedIn