The biggest mistake making your first data hire: not interviewing for product

The number one piece of advice I give startups starting a data team

A few years ago, I met with a struggling data team to share notes on how data teams should operate inside of a startup. There were five or six PhDs with degrees in machine learning and other fields but without a ton of industry experience. The team was having a hard time translating their seemingly impressive data work into material impact on the organization.

Don’t be this startup. There’s nothing wrong with PhDs (of course!), but it seems this company prioritized building out an R&D bench to try to fulfill marketing language around advanced data capabilities rather than being able to connect the credentials with the business. Hiring for product skills early on the data team will help you bridge the gap between data and business value.

Product managers and founders are used to working with engineering teams, but have a harder time translating the company’s vision into something that can be executed on by the data team. It’s important for early data teams to be able to drive their own work.

This may be pretty obvious, but to get a product-oriented data scientist, you need to to interview for product skills. Like — actually give them a product management case study interview. Get a sense for how they prioritize in an interdisciplinary setting and how they connect technical work to the business. Should we spend time organizing data into a data warehouse first, or jump into running experiments?

This probably generalizes to other startup roles as well- so maybe this is generic advice: it is useful to hire entrepreneurial, product-oriented people early on. To me it feels even more important for data teams for a couple of reasons:

• Data as a function is still evolving, so the expectations might be hard to set with this person (setting expectations for the early hires is a topic worth expanding on)
• Similarly, because data as a discipline can be pretty broad, there are a lot of options for how the data team can spend your time. Having awareness of the different sub-disciplines of data and how to sequence them and prioritize them as the company’s mission evolves is pretty critical.

This sense for value will help this person be strategic in fleshing out the team with you over time, and minimizes your risk of having a uniform team that can’t adapt to your company’s shifting needs.

I’m current working on Glean — a new way for teams to explore and share data in— check us out https://glean.io