This is the English version of an article I wrote on my now-defunct German Webanalyse auf Deutsch blog. It is very slightly off topic, I guess, but maybe you’ll learn why great Analytics teams always seem to contain this one total douche, who doesn’t even know what a KPI is.
I read my old post while researching (yes, I actually sometimes google things! Sometimes, I end up on articles I have written myself some time ago. And yes, I sometimes don’t realise immediately. That is one of the reasons why I do this.), and it struck me that four years later, we are in the exact same spot.
Let’s start with the purpose of Analytics in general.
We want analytics to give us data, information, insights, call it what you want, anything that helps us find areas where we could improve the way we work (sell, write, …), and anything that helps us judge whether what we are changing helps, or harms our business. So we want it to tell us where to make changes, and to help us decide whether we keep changes or revert them.
So that’s the job of an Analytics team, stripped down to the core.
It sounds so simple that we all wonder why people seem to struggle. How come we don’t do this, really?
Or, to phrase it positively: what can we do to make it happen? What skills do we need?
I think an Analytics team must have three core skills.
First — Data
Everything we do is about data.
Without data, we’re nothing. Bad data makes us look bad. And even the best data doesn’t always tell us anything useful.
We need someone who loves data, loves tinkering with it, working it, massaging it, until it feels all relaxed, happy, and warm, and starts to divulge its deepest secrets.
That role is often called “data scientist”, and people doing it have PhDs in statistics, economics, or 12 years experience working with SPSS or R.
Good so far? Ok.
That person is responsible for digging out the important stuff, which means the person must understand what is important to their co-workers. The cliché of the person I described above doesn’t always fit that description.
The co-workers often don’t quite know themselves, either, and the data person must therefore be able to guide them, help them. Empathy, pretty important, especially when the subject is data, dry and tasteless.
Data scientists are expensive, the one I described even more so. But they are worth it!
Second — Code
There is no “implementation phase” in Analytics, as we all know. Instead, there is constant change, tinkering, optimisation, and the odd A/B test.
It’s hard to find such a person!
“Let’s hire a proper front-end developer!” you say, and I wager that as soon as you tell your candidate what her job will be, she’ll lose interest.
In my experience, people who do this role are either Analysts who have taught themselves out of necessity, or old ex-programmers like yours truly, who get excited about the Raspberry Pi 4 even though they sometimes find Raspis in drawers that they had forgotten (yes, I probably do have enough of them) (I will still get at least one new one).
Third — Politics
Now this one might hurt a little: data doesn’t/don’t convince anyone.
Information doesn’t, either. Even a killer argument might only look like a killer argument to you, and someone else might not see it at all.
I effortlessly did maths and physics at school. I used to be a developer (C!), and I think of myself as pretty fact-driven. If I see a good argument, and someone else doesn’t follow, I have to tell myself they’re not stupid. They just look at things differently. Yes, that is possible.
So what do you do as an Analytics team when you have presented data, shown analysis, argued your case, and still noone follows?
Well, you need a politician.
Their job is to speak with other stakeholders, find out what is possible, shield the Analytics team from the biggest threats, and advocate for the big things they unearth.
I know what you’re thinking, and you’re right: that person will always be the odd one out on the Analytics team.
The team will treat that person with a mix of contempt and envy. “He doesn’t even know what ‘standard deviation’ is!”, or “yeah, meetings all day, what a tough job that must be…”
But the fact is: an Analytics team has more impact with a politician.
Often, the team manager would take that role.
Also: if you’re the data type, or the coding type, think about how maybe, it makes sense to work for a team where you do not immediately connect with everyone because they think technical, like you do. Think about how a politician can make it easier for you, because they deal with all the human stuff.
As with all rules, there are exceptions.
You might have heard of, or even worked with, a star. Such a person might be able to pull off the combination of all three roles. People like that are incredibly rare, and sometimes incredibly expensive.
Management buy-in might also help, and sometimes it means you need no politician. In that particular case, I would recommend finding one, anyway. Who knows…
In the overwhelming majority of cases, an Analytics team should come with at least two people: a politician, and a doer. Three would be better, and obviously the more, the merrier.
7 thoughts on “Analytics Team Roles”
First: great article. It resonates a lot in this period of my career.
“Well, you need a politician”.
In my opinion _politician_ may raise mixed feelings in the reader. In my experience almost all projects, not only Analytics related, need someone who advocate the project and mediate across different stakeholders. It that a politician role? Never thought in that term.
A sponsor is trusted and acknowledged, he inspires the other stakeholders, creates consent, empowers the team.
That does not exactly my idea of a _politician_.
I should have defined the word, I guess, or rather what I meant when I wrote “politician”: I am thinking of a person who gets input from all stakeholders, then tries to find a way forward. Very simplistic, but it is extremely complex because we’re all humans.
The “politician” would be listening, aggregating, evangelising, discussing, bartering, negotiating, talking, talking, talking. They would also handle internal PR, in a sense, make the team look good, and deflect any threats.
Fact is that often, neither data nor code person are very good at _using_ the results to incite changes.
Would love to hear your thoughts on how to build a strong relationship between your Analytics team and your IT group. So many Analytics people I’ve known over the years simply roll their eyes when IT is mentioned, and honestly many/most IT people do the same about Analytics groups. What have you seen work for different organizations where there has been a healthy and productive relationship between the two?
That is a big topic… and I haven’t really seen anyone who made it work, really work.
We have spoken about one aspect of it, a hard, well-defined interface between the two teams, and I think that is one important building block. But it’s not enough, on its own.
To be honest, I don’t know either.