Thanks to a wealth of online courses (Coursera, DataCamp, EdX, and many more), there are many ways you can become comfortable working with data for little to no cost. Developing these skills is an important step in becoming a savvy, innovative, and respected PM. The decisions you make from discovery to market and everything in between can be informed by data and the more you understand its powers and limitations the better positioned you’ll be to make smart decisions.
Beyond the ability to write and run queries, analyze trends, and build reports to track product goals, learning the fundamentals of data science will strengthen your ability to know what to ask of the data. And this — the ability to be curious, strategic, and clear — is what can set you apart as a partner to the data science team.
As a bonus, you’ll also learn parts of the developer workflow (e.g. how to work with git, efficiently browse sites like stack overflow, and work with the command console) as well as basics in data visualization — all of which are important parts of an advanced PM’s toolkit.
Every data science team is a bit different. Sometimes they operate as their own unit — fielding requests, conducting research, and building and tending their data lake. Other times, individuals from the team act more like an engineer on a product team, working side by side with the developers, designers, and PM to innovate, iterate, and bring products to market.
However your Data Science team operates, you should ask them about their practices, priorities, and interests, and learn how they would ideally like to work with you.
Whether you’re working with them to implement an eventing strategy or develop a larger efficacy study, knowing how they want to be involved in each stage of product discovery and development will help your relationship and develop trust.
Take the time to learn what systems and tools they use — and why. Sometimes it may make sense for you to run your own analyses and build your own reports, but for some situations and teams, that’s a real trust-breaker.
Data Scientists excel at exploration, experimentation, and rigorous measurement. The way to partner with them to optimize these strengths begins by asking questions you want to answer about your users and product.
It is definitely within your purview as a PM to identify key metrics for your product from acquisition to engagement and retention. There are many frameworks and resources for getting started on this path. In addition to the great guidance out there, I’d add one I don’t really see: don’t do it alone. The data you’re collecting for your product KPIs likely only scratch the surface for what data scientists are capable of looking at with a wider view, more data, and over longer periods of time.
Let’s make this concrete. Say you are tracking goal conversion rates for 4 user segments and notice that segment 2 out-converts segment 3 by 30% over the last 2 quarters. At this point there are a number of things you can do — you can look at the cost of acquisition for each of those segments to see whether that additional conversion percentage is actually good for your bottom line. You can explore the different marketing channels used to see what campaigns have brought in each segment and how much that might account for the difference. You can do most of this yourself if not with a little help from a product marketing manager.
But you can also ask a bigger question working with your data scientists: who are our best customers over the long term and why?
There are ways you could do some of this analysis yourself. Where the data scientist toolset comes into play — and where working with data scientists who understand your product, company, and market well can give you major leverage — is in being able to break down that meta-question into discrete analytical steps that might just result in breakthrough insights.
Building from the example above you may find that the way you’ve been thinking about your segments is entirely wrong or that while some variables that you add solid short-term results actually hurt your longer-term profit. These kinds of insights are really difficult to parse out when as a PM you’re tracking results on a tighter timeline and typically come to any analysis with some preconceptions and biases about your product.