Do you know if your work is going smoothly?

How can you tell? And what is smooth, anyway?

How is the quality of your work? And maybe more importantly: has it improved or worsened? Clearly this is of huge importance to your business; if you aren’t able to deliver decent work to your customers, word will get around. People will say that you used to be pretty good, but now you’re slacking off.

So if you get just one thing right, get this right!

Everyone has a gut feeling on how things are going, but how can we be sure we are right?

That’s how: objectively measuring things, and drawing conclusions from these measurements.

CMMI likes to call this Measurement and Analysis. It has this to say to introduce the subject:

“The purpose of Measurement and Analysis (MA) is to develop and sustain a measurement capability used to support management information needs.”

And while they’re right in spotting its importance, their ideas tend to get lost in big words and corporate speak. So let’s try to untangle them:

There are three steps to Measurement and Analysis:

• Finding things you want to measure
• Collect and store the data
• Analysing and understanding it

Step one: What is worth measuring?

That’s the big one.

It very much depends on your project and your organisation, which is why concrete suggestions are out of place here. But there’s one general rule: there’s no point in measuring things you have no control over. While that may sound obvious, it’s surprisingly easy to fall into the trap of agonizing over issues that are outside your scope of influence.

If you don’t have the resources to improve a certain issue, what’s the point of tracking it? It’s going to be the way it is anyway. Same thing if it’s outside your control.

So, spot things that

• you have control over
• you care about
• are measurable

Also, while you are starting to measure, don’t go overboard. Start small, and once you grow more experienced with this tool, expand your focus. Keep in mind that measurements quickly develop a life of their own. They demand your attention, and subsequent pondering as to their meaning and consequences. So be careful not to swamp yourself in potential topics ( They might be worthwhile. But perhaps not now.).

So, as a practical example: how about the number of bug reports from users? That might be a nice metric to track, and attempt to improve.

Did you notice the emphasis I placed on users? Think about other sources of bugs you might want to track: internal testing, for instance. Tracking these may not a bad idea either – but it would be easy to improve your numbers: just test less. Clearly, that’s not a smart move.

So pay attention that you are measuring the right things; it is very easy to set wrong incentives, and ultimately achieve undesired results.

Step two: How can you know things?

Now that you’ve identified what you want to know, how can you measure it?

The answer, again, is a resounding “it depends”. Try to spot quantifiable factors which influence the thing you are interested in.

However, you should always strive to automate the measuring process, or otherwise make it as effortless as possible.

If it’s cumbersome, it will hinder your productivity. And: it won’t get done. It will be put off, or corners will be cut. In any case, you won’t be able to reliably gain the insight you are aiming for.

To stay with our previous example: use your bug tracker (you do use one, don’t you?) to count all new bugs of the last week, or the last sprint, or some other suitable time frame.

Do this regularly, and record the results.

Step three: Now that you know, what do you do?

After you’ve collected some data, try to analyse it, and derive decisions.

It frequently helps to display it as a graph.

How can you improve the situation the metric points to? Note that I didn’t say improve the metric? I can’t stress this enough – chasing numbers is almost always the wrong thing to do. And if you know that there is improvement, but your metrics don’t show it, you’re collecting the wrong metrics.

Always keep the old adage in mind: measure twice, cut once. Be really sure of what your data tells you, and what your conclusions should be.