Tuesday, June 9, 2020

Data don’t generate theory – only researchers do that. (Henry Mintzberg)

Data doesn’t speak for itself. In fact, data can be sometimes used like a ventriloquist’s dummy – parroting whatever the ventriloquist wants to say. It takes someone with some real skill to get the data to really talk. And a lot of hard work as well.

Now, data alone will tell you that something did indeed happen. That, though, is kind of like learning that your car won’t go. Well, why won’t it go? What can you do to make it go again?

So, say the completion rate on your account opening process is below 10%. That’s good to know. Doesn’t sound too good either, does it? So, what would actually be your next step here?

Well, one thing would be to dig even further into the data. You might, for example, find out that certain users have less trouble than others. You might also find out that most people drop out on step x. Who knows, there might be even certain times of day when people are more or less successful. Honestly, though, your options will be limited. Web analytics are great for straight-up, high-level numbers. What those actually mean when you get down to it, however, can be another thing altogether.

Another possibility – and a very popular one I might be add – would be conjecture. Heck, that’s all A/B testing is when you really get down it.

A third idea, though, would be to get better data. And that’s why I’m always pressing for qualitative research. Usability tests, ethnography, in-depth interviews, even focus groups can get at some of those thorny why questions. 

In fact, good qualitative data, along with some serious analysis, will start to get at why’s that don’t apply just to the particular project you’re working on, but to multiple projects over time, and in multiple different situations. Throw in a little more explication, maybe a metaphor, and – and hey presto – you’ve got yourself a theory! 


Dang! – my kind of management consultant

No comments:

Post a Comment