Grounding Numbers

There are a bunch of different types of work that are broadly called “data science” but today I’d like to talk about insights. For me, insights work is all about changing minds. Perhaps ‘changing’ is too strong and adversarial a term: if we give a decision maker extra confidence, that’s probably still OK. But the gold-standard of a good insight is that it leads to somebody changing what they were going to do.

It’s much less certain than a bunch of other data work because it deeply involves people and people are deeply unpredictable. If you can sit and work closely with the person you’re trying to influence, it’s much easier to understand what they’re motivated by, what they believe the key facts are, and then how you can fit into their process and offer them value.

But so much of very senior insights work isn’t like that at all, and much more closely resembles research work in nature. You spend a while working on an area and building expertise, then you take a chance. You find something that you think is interesting and should change minds, but are you right? Have you distilled it perfectly into just the right bite-sized chunk? Have you packaged it in just the right way? And then when you launch it, do you hit the right people at the right time?

In my experience, senior executives have a few key grounding numbers: snippets that help shape (and explain) their position on topics.

“I know that the average person on Facebook has 372 friends”

or, every data scientist’s least favourite example:

“People with 10 friends within their first 24 hours are 75% more likely to be retained”

Good insights-based data science work is creating those snippets and getting them to land. I imagine it’s fairly similar to a lot of work in politics, where you’re trying to get a slogan or a data point out into the general consciousness.

However, even after all that work, and even if you got everything right, you can still sometimes miss the mark. Maybe your stakeholder’s priorities have shifted, or they’re getting pressure from elsewhere that they can’t ignore, or they didn’t even read what you shared. You necessarily have less information than them, and so while you might think your thing is critically important, they might disagree.

If all of this doesn’t sound like data work…maybe that’s because it’s not. At senior levels across so many careers, influence becomes the mark of success and your skillset is just the tool you lean upon to have that influence. A senior engineer, designer, user researcher, or data scientist at a tech company has to be technically strong. But they need to know how to have an impact as well, and having an impact means being able to influence whoever makes the decisions. I once read somewhere that by the time you’re a “Senior X” you should be able to pick up anything technically required. To be Staff+, it’s all about how you manage to apply it.

Anyway, I just launched a bit of work and I’ve got no idea if it’s going to land or not. I agonised over it for a few days and polished it as best I could, but now it’s in the lap of the Gods. God-speed, little insight. Land well, and let’s see what tomorrow brings.