In 2015, I wrote that I think “agile analytics” would currently not work, mainly because we lack people who take on the “product owner” role, i.e. people who are good at guiding, prioritising, evangelising, and occasionally steamrolling, our efforts.
A year later, I opined that as an industry, we are not a great fit for agile, but that we had to be aware of agile, and, crucially, work within the bounds of it to be successful. Mainly, since we needed our developers to an extent, we needed to understand how they (you) work.
In 2019, I came to the conclusion that for Analytics to matter, you need a politician. A politician in the raw sense of the word: someone who manages to get people to buy into an idea, someone who motivates, communicates a vision, and makes sure people know where to go.
I think that the best approach to useful and sustainable analytics is to start small, learn, build up, and never stop evolving. There is no way you can know now what will be important next year. Think about all your plans from 2019. I’m guessing a lot of them are quite obsolete…
Because of all of the above, I think we are pretty well placed to work with anyone who works agile, even if we might not use agile principles, ourselves.
From the point of view of an agile development team, we should be easy to work with.
We are happy with short sprints.
We love being able to reassess things.
Most people in Analytics think that data will lead to changes, or to decisions. We are totally happy with a plan that adapts to new insights. So, if an agile project might not absolutely build what was planned, but instead build what makes sense, we are happy to be a part of figuring out what it is that makes sense.
We should be able to write a good story.
This is a funny one. A lot of people in our industry, including yours truly, believe that measuring should always have a purpose. Ideally, someone would voice a theory, and Analytics would provide the data needed to prove or debunk that theory. That approach sounds like a perfect agile match. But, alas, we’re not quite there, yet.
A corollary: We should be able to write tests.
You all remember how much I was writing about testing in Analytics. I haven’t given up! And how cool would it be if we, the analytics people, did write tests, so you, the developers, could implement better?!
Makes me hope, though, that the more Analytics people work with agile teams, the more we shall write stories, and at some point write tests, too!
The other way round, from our point of view, agile development teams are also great.
You speak with us, often. (As opposed to the usual “tomorrow is go-live, can you guys please add Analytics, today?!”)
You are open to hearing things like “well, we measured that, and it doesn’t look that good”, and then acting on it.
On a completely anecdotal level, you are also great to work with!
The few times that I had the pleasure to work with proper agile dev teams, it has been so good! Easy communications, a willingness to figure things out, openness on both ends, and some of you have a drive that simply infects us, and makes us happy!
Dangers & Opportunities
One thing that hasn’t changed in the time since 2015: a lot of the projects we work in, especially the bigger ones, are anything but agile.
I notice that, because contrasting the “agile” with the “standard” projects is just sad. It hurts.
And there are still people out there who think that this whole digital transformation thing can be done in one, single, big effort, “and then we’re done, right?”
But implementation is not key, and it makes sense to build for change, in the sense that people, process, budget, operations all must take into account that “this digital thing” changes all the time!
What the agile mind frame brings, and what our trade really needs, is the realisation that nothing is static, all targets are moving, and that that is a good thing!
Our optimisation colleagues know that. They are living it, and so should we.
2 thoughts on “Agile & Analytics – still Friends”
Ah there’s so much I relate to here. Personally being in an agile transformation role, these things take a long time indeed. Particularly tying in stories, definition of done, test cases and OKRs. Gosh, I completely hear you about the process by the time all of these tie together. Personally my goal is to tie the OKRs into an agile process, but they all take such a long time