I’ve been fortunate to have had the opportunity to grow the data team at Lookout from one person (myself) to a team of outstanding analysts that drive insights across product, marketing, engineering, finance, security, and support. It’s been a long road. It’s taken two years, in which time I’ve probably interviewed (via phone and in-person) over 50 folks. If you’ve ever been in the position to hire data analysts, or been responsible for their performance, then you already know the obvious…
Hiring great data analysts is hard
A director of analytics at a startup once told me that he was batting .500 on his hires, and that he thought that was actually pretty good. That is good. I believe that the two hardest jobs right now to source in the tech world are product designers and data analysts. There are lots of reasons why great data analysts are hard to find, but here’s a short list:
- There aren’t great signals. There is no “data analyst” major at UC Berkeley. There is no one technology that, if on a candidates resume with corresponding label of “expert”, guarantees them a minimum level of success and competence in the field.
- The data analyst field is new. There just isn’t much employment history there. You don’t see a lot of people floating around who have, let’s say, 5 years of experience as a product data analyst at XYZ Hot Growth Company. What this means is that 95% of candidates that you look at will be transitioning either from a tangential role or completely switching careers.
- There is no one skill set you can test to guarantee success. Not that you can really do this with any hire. But, at least with some roles there are basic tests (code challenges for engineers, portfolio reviews for designers, etc).
- If they’re experienced, they’re probably not looking. Chances are, if they’re good and they’ve been doing data for a while, they’re likely very happy. Which probably means that you’re not going to poach them unless you happen to drop that LinkedIn InMail on their lap just when they happened to have an incredibly shitty day.
Generally what this means is that you have to be open minded and look at a lot of different folks, with a lot of diverse backgrounds, all of whom [should] have a passion for data. That’s a big challenge. Here’s what I do to make sure we get the best analysts:
1. Look for intellectual curiosity
Intellectual curiosity goes a long way in almost every job, but it’s pretty much table stakes for a data analyst. Upon uncovering an inconsistency, weird hiccup, or peculiarity in the data, a data analyst must have the innate curiosity to want to double click into it. I can’t tell you how many amazing insights we’ve found as a team that way. If an analyst is looking past the data and just focusing on getting the analysis done, they’re missing the point and you’re missing business opportunities.
2. Look for tangential analytical skills
As mentioned above, finding DA’s with direct experience is generally very difficult. Not to say it doesn’t happen, but in my experience, it’s been more the exception than the rule. That means you have to find other ways to determine how analytical the candidate is. This doesn’t have to be something super comprehensive (i.e. I’m not looking for someone to have written a research paper where they dissected voting patterns across rural areas in blue states). But, I want to know that this isn’t their first rodeo when it comes to digging into a data oriented problem.
3. Look for the ability to think critically about problems (in a data driven way)
I love giving case interviews. I find them to be the absolutely best predictor of success for data analysts. When I give them, I don’t have a right or wrong answer (I know, I know … people who give case interviews always say that, but I actually mean it). What I want to see is how a candidate can take a nebulous business problem and apply data constructs to it to produce actionable insight or follow up analyses. Or in other words, I’m looking to see if they can do the job Being great with SQL, being analytical, being intellectually curious, and so forth – those are all great – but a great DA needs to understand how to add structure to a problem using data.
4. Look for them to grasp basic concepts of transforming and relating data
Given a straw man data set, how quickly can a candidate pick up on how data conceptually relates to one other? That’s pretty much it. I try not complicate this and I generally only spend a few minutes on it. I just want to see if someone can understand the basics of how certain data relate. If they can’t get this, it’s generally a huge red flag.
5. Look for them to move past analysis and onto business insights and actions
This is the hardest skill to find in a great data analyst and it’s what truly separates the best from the rest of the pack. A great data analyst takes their amazing analysis and goes one step further to tie it into some business decision, recommendation, or follow up action. You have $3.75 12-Month ARPU and a $3.25 CPA, great. What do we do that? Is that good? Should we look at LTV? What payback period are we comfortable with and what is the certainty on the cashflows? Given these considerations, are we at profitable acquisition? The data analyst needs to own and drive the analysis, not just be the cruncher of the stats. The ability to move from numbers to true insight is a rare gift (albeit one that can definitely be mentored and developed).
And what not to ask…
All of this said, I’ve also found a few things that tend to be red herrings for predicting great analysts. Here what I generally don’t consider:
1. Don’t look for domain experience
At Lookout, we hire athletes. Our analysts switch between marketing, product, support, engineering, security, and so on. Make no mistake, it’s very hard to hire great athletes. For starters, they’re rare. Secondly, the things that you’re looking to test them on are incredibly difficult to tease out in the interview process. That said, it’s 100% worth it. A great athlete can solve complex problems that transgress data silos and organizational teams. And, they unlock incredible business value in doing so. The reality is that most data problems are basic math. Things like CPA, ARPU, conversion rate, and retention rates are just definitions. Picking up and understanding domain specific concepts like these is straightforward for great analysts with solid foundational skills. Domain experience just isn’t that important.
2. Don’t look for SQL experience
I confess: I used to get very caught up in SQL experience. Probably 90% of all analyses at Lookout starts with writing some flavor of structural query language syntax (either HiveQL or MySQL). Given how much we use those environments, I previously believed that having a deep skill set in SQL was imperative to success. I couldn’t have been more wrong. The reality is that SQL is easy. It’s an intuitive language. With the right training, most people pick up the basic concepts and are fluent in simple queries, joins, and aggregations in less than 30 days. Don’t optimize your hiring for the first month when it’s long term performance that matters. Hire great people who can think critically about problems in a data oriented way and teach them the basics to munge data.