‘We now know that our project is going to cost £1,245,873.55. Great work!’
‘And if that means booking two hours for John in December 2016, then so be it. Remember, this project will only get it right if we know exactly how much time is needed, who needs the time, and when they need the time to be scheduled.’[Note: this was written in Q3 2015.]
‘Only if we know the estimate can we know whether or not to escalate. You know how it works—more than 10% out, and you have to go to the steering committee. Your project is then on red.’
‘And we need the estimates so that we can compare the abilities from project-manager to project-manager, you know what I mean, and then we can get rid of the rotten PMs.’
If you think Agile fixes situations like this, how wrong you are!
You think I’m going to be writing about something you already know about. It is well documented that during the 70s, 80s, 90s, and 2000s, we thought we could predict projects. We simply lied to ourselves, saying that we could control them. We thought that if we knew the estimates, they could be our baseline against which to measure everything.
Complex governance was needed in IT departments, which was in line with the then-current standards and best practices of the time.
Then, around 2010, along came Agile and story points into the mainstream and things actually got worse! IT stopped wanting to make high-level plans, to the horror of the customer. The result was a backlash from management, who wanted to throw out Agile.
Do You Have This in XXL?
It is also quite frightening to think about the amount of research and effort that has gone into different estimating techniques. T-shirt sizing, min-max, most likely, etc., all of which were then fed into a project management tool to calculate the critical path.
All of this gave us a false sense of security.
But it is not the estimating techniques themselves that are to blame. It is those who have decision-making power, the control freaks who make us stick to exact estimates.
Are You Satisfied?
But first, I want to talk about something else. I want to talk about ‘satisficing’ and what could possibly be the most powerful word in the English language: ‘reasonable’.
I once had an ex who I discovered, while shopping for skin cream, had sat down cross-legged in the middle of an aisle in a busy shop on Saturday afternoon. She had opened five pots of skin cream, and was testing all of them, but she couldn’t make a decision despite the fact that the price difference between the most expensive and the cheapest could only have been 80p, maximum.
She could not decide. The decision was too overwhelming for her. She couldn’t accept the consequences of possibly choosing the wrong one.
Yet the ability to just make a decision that will be satisfactory is what is important. We can do it quickly.
Herbert Simon was a Nobel Prize winner and one of the founders of the fields of organisation theory and information processing. Simon coined a term for deciding on something that is simply enough for our needs: ‘satisficing’.
In other words, I might not have all the information I need to make the best decision, but I’ve at least got enough to make a satisfactory decision.
And we do it all the time in everyday life: cleaning the house, travelling second class instead of first for that 10-minute train journey, going to a non-brand store to buy underwear, buying a non-Apple phone charger, and so on.
We Are Scared
Yet when it comes to estimates, we don’t satisfice. We are scared.
So why don’t we practice satisficing with estimates?
How much time do we spend trying to determine the exact price of a project when it will all change anyway? Isn’t it highly likely that all the effort we’ve spent will be wasted? John may not be working with the company in a couple months’ time, let alone in December 2016.
15 Years behind Bars
It is certainly not reasonable to expect a project to evolve exactly as we thought.
Aha! There’s that word—‘reasonable’. In English law we use the expression ‘beyond reasonable doubt’. It has probably sent more people to jail and set more free than any other phrase. But very few people have gone to jail for an incorrect estimate. [Okay, some may have, especially in the lawyer-filled USA.]
Is it really reasonable to expect somebody to find out exactly how much the project will cost months in advance of delivery?
The question we should be asking is more like this: is the effort required for this project reasonable? In other words, does it make sense?
By asking this question, we have to rely on our judgement, as well as analysis.
But we need to get used to using our judgement.
We have to use both sides of our brain.
What Can Agile Projects Do?
Clearly the best way to do this is to provide the range at the beginning of a project and then constantly manage it.
Of course, at the beginning, the range is very imprecise and wide. But that is reasonable. It also allows us to decide quickly whether or not to progress.
It helps our satisficing. We should not need to sit on the floor, but let’s face it—sometimes those steering committee meetings make you want to!
Satisficing allows us to make a judgement call based on an order of magnitude. Is my new website going to cost me £10-20,000 or £150-200,000? This is all I need to know to make the decision to progress or not. And believe me, getting that wide range is very quick and cheap.
In other words, the extra effort and cost to find out that the project is going to cost exactly £18,782 is just not worth it. It’s not reasonable. It’s just a waste of time. Why? Because that estimate will change.
And you know what? Knowing a specific estimate isn’t very satisfying either!
Still not convinced? You know darn well that as time progresses our estimates will improve because our knowledge of the project becomes more precise.
What is satisfying about that? Well, think about it. At the time the decision was made, there was some risk, some unknown—yet the decisions were made. In the end, your judgement call turned out to be right! Now that is really satisfying!
How to Be Satisfied
Notice that I’m constantly looking at my estimates. I don’t do some terms of reference with a business plan or project plan and fix myself to it. No, I constantly look to ensure that my project is making sense. I ask, is it still reasonable to progress?
So how does all this relate to Agile techniques?
Well, in Scrum, we don’t have estimates beyond story points. If we are going to do a rather large project, the team has certainly not assigned points to all the stories. In fact, many stories will not even have been written. This is one of the reasons why a Scrum team is useful for continuous product development but not necessarily for projects.
Kanban involves no estimating.
Both Scrum and Kanban support satisficing at a very low level, with story points on user stories. They rely more on the statistics of flow to know how much has already been done and how much must still be done and they extrapolate from there. The concepts are very simple to understand, but the realities are far trickier. For one, you need a stable team and environment—this alone is difficult to achieve.
The portfolio level of SAFe is supposed to provide guidance with Epics, estimating, and business cases, yet SAFe provides little guidance on exactly how to do that. You still need to know the high-level scope of your project. SAFe is not designed for this. Dean Leffingwell himself says that SAFe is intended to be used for flow.
However, DSDM as a project management framework not only supports satisficing, but actually requires it: the estimates get more and more refined as time progresses and as the project moves closer and closer to completion. There are constant checks to see if the business case makes sense.
If you need to have estimates prepared for an entire project due to your culture, history, and governance, then remember the theory of satisficing to judge whether it is reasonable to progress with a project. Don’t waste time trying to get an exact number—it will change anyway.
And You? Have a Think about These Questions. (Book 15 minutes if you want to discuss)
- When was the last time you demanded an exact estimate?
- When did someone last ask you to deliver an exact estimate?
- Was that estimate reasonable?
- How could you have lived with the decision to progress? Could you have satisficed?