When Joakim and I wrote the book we had a chapter on measurement in it, chapter 11 - “Using metrics to guide improvements”. It was intended to show a few ways that metrics can be used in a flow-based process that uses kanban for improvements.
When we wrote it I happened to show it to Torbjörn Gyllebring since he’s very sincere in his criticism. His first words:
You can't write a word about measurements if you haven't read "How to measure anything"
When you have not read that book and writing a lot of words on measurements… hearing that has a bit of a “DOH!”-effect on you and your writing. But Joakim had and that made me feel a little better. I was largely satisfied with the chapter too.
But now I have read it and I wanted to share some of the main points that I’ve picked up from this great book by Douglas W. Hubbard. How to Measure Anything: Finding the Value of Intangibles in Business is the name and link for the book. It’s an awesome read and I recommend anyone to read it.
Everything can be measure or else…
The first thing found profound is that even intangible things can be measure. The whole book starts with that statement in fact:
Anything can be measured
(Not saying that everything should though…) But quite often I’ve ran into situations and people that says that “our services provides a value that cannot really be seen or measured”. I was therefor happy to see that Mr Hubbard list such a situation too in the second chapter, in a sidebar on the Mitre Information Infrastructure (MII) whose CIO said
Our most important gain can't be as easily measured - the quality and innovation in our solutions that becomes realizable when you have all this information at your fingertips
Now if it doesn’t make a measurable effect - why does it matter at all then? I think that this often has to do with us not being able to think up new ways of measuring the effects. Because, as Mr Hubbard states a couple of pages earlier,
even touchy-feely-sounding things like "employee empowerment", "creativity", or "strategic alignment" must have observable consequences _if they matter at all_
Luckily the rest of the book is devoted to helping us finding and understanding ways to measure things that we didn’t think could be measured.
To me this has a profound effect on what we are actually measuring, maybe even what we are doing? If we are doing doesn’t make any observable difference - is it worth doing? Or can we restate what we are doing?
Decisions, decisions, decisions
Prior to making a measurement, we need to answer the following: - What is the decision this measurement is supposed to support?
This really the reverse of the paragraph above - what are we going to use this measurement for?
I am the biggest sinner here, so I’m happy to talk about this freely. Very often I measure things because they are easy to measure and makes for nice information, rather than thinking about what I am going to use the information first.
Let me take two examples that is commonly used in agile teams with visualized flows on boards, and that we suggest in the book too:
Lead time - how long time is each item on the board? Easily measured by simply writing the date on the sticky as we hang it on the board. We it’s completed we note that date. Subtracting the two gives us the lead time.
Throughput - how many items gets Done each week. Also trivial to measure: each Friday count the number of items in the Done column. Write it down. Remove the sticky.
There! Easy to measure, easy to catch. But useful? Maybe. “What is the decision this measurement is supposed to support?”
Often I have no answer to that? Or at least, often there’s other more interesting things to measure. So, our lead time is short - but are we producing quality? Ok - we have a great throughput - but is the work environment healthy, or are our customers using the things we are putting out there with such high speed?
What, really, is the decisions we want to affect with this data? In the book Mr Hubbard also tells the story about a bank putting a lot of effort in getting reports from their branch offices, but couldn’t recall a single instance where those report had changed their decision. Talk about waste. That effort could have been used better, I’m sure everyone involved agree on.
Clone the world
Mr Hubbard proposes a thought experiment to help you find what to measure:
Imagine you are an alien scientist who can clone, not just sheep or even people but entire organizations. [...] What do you imagine you would actually observe-in any way, directly or indirectly - that would change for the first organization?
This is an awesome way to get to the bottom of what to measure or at least starting to think about what is important to measure. In fact, I have used an variant of this for a while, no idea what I’ve picked it up:
If I wasn't here - how would you know?
For example; an pastor at a workplace - what should define her role? One way to start is to ask; if we had no pastor at this workplace; how could we notice the difference? After 1 year? Each day? In staff? In our customer? In the way we do business?
By the way you can ask the same thing for many supporting roles in “normal” workplaces; agile coach, managers, HR department, IT-support to mention a few that I’ve been involved in.
How to measure anything is a great book and I’m happy that I’ve read it. I was surprised though that the main points that it gave me reached far beyond just measuring things:
- If we are providing a service “which values cannot easily be measured” - maybe we should think again about what we are trying to achieve. Some kind of observable consequence must be present if they matter at all? Right?
- Measuring things just because they are easy to measure is pretty useless. What are the decisions you want to support? Will this measurement help this?
- One thought experiment to find what is important to measure is “if we didn’t do this - how would you notice the difference?”