A Data Product Managers Take On Hiring Data Professionals
How does hiring Data Professionals Work at a Startup. A must read if you are struggling in the interviews.
For this post, we have an actual hiring manager who works at a startup, and he has had to adapt quickly and hire an entire data team. If you are struggling in your interviews, this post is a solid read on what the hiring manager’s mindset is like.
This entire post was created by Bowtiedwhitebelt
You can subscribe to their stack here:
Intro
As can happen at startups, reporting structures can get weird. My startup is no exception, through a confluence of events, I now have an entire team of data analysts reporting to me as the data product lead. When I first started hiring said data analysts, I had a recruiter who couldn’t tell the difference between and analyst and scientist, and a massive knowledge gap to cover in being I’m a dumb product manager, and little understanding of what data analysts in the marketplace actually want in a job.
Having only hired product managers in the past, I knew I needed to level up quickly.
So - I did what any smart person should do - I looked to the jungle.
That is where I first connected with BowTied_Raptor - who helped me understand what the hell a data analyst actually wants in a job. We went back and forth and I was immediately able to identify my issues in hiring. (Check out Refining You Resume by Raptor for some more details on this front)
Who am I? I’m bowtiedwhitebelt, your local expert on all things product management. I’ve specifically been tasked to build a Data Product & Analytics org from the ground up! What a long, strange trip its been!
Issue of a Hiring Manager
One issue I’ve found with hiring data professionals is that Data Science lives on a spectrum. At a startup, this presents issues due to the nature of the work being all over the place. Sometimes I just need someone to run a join in SQL to combine two tables for a report some ad hoc insight. Other times I need predictive algorithm based based of a data set assembled from multiple scraped sources.
When I started hiring, I had basically anything and everything you could possibly do in the Data Analytics team list in the JD. This resulted in a very confused recruiter - as he kept trying to bring in far more senior candidates to cover everything as opposed to the right person with the right experience to execute the specific tasks at hand.
Here, Raptor was instrumental. He gave me the suggestion to break the JD into 4 separate sections.
Day to day work - Where I explicitly highlighted the most common projects I would assign the hire
Expected skillset - From most important to least important, what you have to be able to do to accomplish the assigned work
Bonus Skillset - This was where I could highlight some of our more difficult problems that would be nice if someone could help with, or specific tools we use that could cut down the learning curve
Why you should want to work with us - Here I sold the benefits of my company to the candidate
Here’s a good job posting:
This helped candidates self select into the role tremendously - reducing the stress levels of both the recruiter and myself. I immediately felt the impact of better fit in interviews.
What We Looker For and How We Vet Candidates
Technical Competence
We knew with a hire we would have to be heavily involved in educating them in our data architecture as part of onboarding. We did not want someone who we would also have to help tighten up their queries and code. We did this with a SQL quiz that had a 25 % pass rate with V1 of the JD, and about 50% pass rate with V2 (Thanks Raptor!)
We wanted SQL so that we could see how candidates would approach unknown data. In practice, we have an infrastructure to query in Python on a day to day, but we wanted to go old school, as our production data all lives in Postgres.
General Understanding of Data
After we established technical competence, the next most important important attribute was how you would approach any new piece of work assigned.
For this, we would use a case study of an example problem we want to solve. This included a business explanation from me, then a sample data of raw data.
What we most wanted to know - are you thinking about the actual process behind the data?
We intentionally added some inconsistent data to table that represented some idiosyncratic business process. Follow-up questions would involve them troubleshooting these inconsistencies. We’re a small team that moves fast, therefore we need people who try and understand what real world factors affect the numbers on the screen. In practice, the best for us to judge was by the quality of the question the candidate asked in the case-study interview. One candidate dove so deep he answered our follow questions before we could even ask (A very good thing). All this can be summed up as “Can you think critically about what you work on?”
Culture / Fit
The last measure we would evaluate in a candidate was a culture / fit. Essentially, your ability to collaborate and play nice with others. This become critically important in our org as everyone on the data analytics takes an outward facing role, meaning our Senior Data Analysts or Data Scientists actually get on our client calls. Mid-level and junior team members then are also involved on internal discovery calls and asked to explain their work to the non-technical teams in the business. To do this affectively, you have to be able to work with others smoothly.
On a personal note, this has been a boon for me as a non-hands on manager. Our data team better understands what needs to be done when they’ve heard it from the source of a request. They then also can get feedback from an SME without having to involve me. (I, being “Leadership” get invited to an ungodly amount of meetings).
Lastly, it enables direct feedback - my team can hear how their work is consumed, as opposed to second hand from me.
What Success Looks Like for a Hire
We try to work on an initiative basis - and our data analytics team is in very hot demand by the company as a whole. We commit individuals each quarter to work on projects based on discovery of the data product team, after official sign-off from the execs.
In practice, this means a successful data analyst must be understanding, analyzing, and delivering actionable insights for any initiative to which they are involved. This includes hitting deadlines. A large part of my day-to-day is scoping down the projects to deliver things quicker. The ability for an analyst to be reliable is huge. Note - this doesn’t necessarily mean “quick”. It means if you say “I can get this done in a week” you then actually get it done in a week.
As management, I’m in the business of forecasting work for delivery. Help me do that better, and I will shower you with raises/promotions/bonuses.
Better yet, if you can tell me how we can execute in quicker, more effective manner, I will start letting you commandeer other members of the team to help you deliver. Demonstrate good judgement, and I will let you make more important decisions. I have too much to do day to day, so a hire I can trust to operate independently is the best possible outcome.