There’s a fine line between “science” and “magic,” but you’d never know that from some of the job descriptions for data scientists. Build systems! Collect data! Analyze pipelines! Write reports! You will be responsible for transforming us into a “data-driven organization” all by yourself! Data can unlock massive growth potential for your company. If you expect one person to successfully wield that power, you’re not looking for a data scientist. You’re looking for a wizard.
What you need is a team. And for most small companies, that team doesn’t need a data scientist or a wizard.
What does a data scientist actually do?
The irony of hiring a data scientist to run engineering and analysis is that neither of those is really in their wheelhouse. A data scientist applies statistical methods to complex data sets. They write algorithms for machine learning.
A growing company with a few million dollars in revenue doesn’t need a data scientist yet. It needs data processes and structures. There’s nothing a full-time data scientist can offer a small company trying to make data-driven marketing and operations decisions.
Data scientists hired for engineering and analysis quickly get bored and leave for greener pastures. If they don’t, they may not have been a true data scientist to begin with.
As misplaced demand for that role rose over the last decade, boot camps and certificate programs sprang up, flooding the market with underqualified people calling themselves “data scientists.”
No matter what you call them, when a single person tries to do it all, the result is the same – an overwhelmed employee hacking together one-off reports. It doesn’t matter how skilled or experienced a person is. There are not enough hours in a day to respond to a constant flood of service tickets and also perform deep analysis.
Who should be on your first data team
The time to hire a data team is earlier than most companies think. As a rule of thumb, if you are doing enough business to support twenty full-time employees, you need someone – even if it’s only a consultant – managing data.
The bigger and more complex your company is, the more complex your data team will be. A startup building its first team should start with two fundamental roles: an engineer and an analyst.
The data engineer
A data engineer knows how to build processes and systems. They will set up your infrastructure and build your data warehouse. Once the foundation is in place, the data engineer maintains the pipeline.
The data engineer reveals how information interacts between your different source applications. If your sales data lives in Shopify and your customer service tickets live in Zendesk, a data engineer is the one who can tie that information together and create a holistic view of your customer. They unlock the ability to scale insights and monitor trends over time.
A good hire knows how to manage a data pipeline and how to work with data warehouse and integration tools.
The data analyst
A data analyst translates the information the engineer provides into business value. They give meaning to raw numbers and show how the information applies to business questions.
While an engineer can get away with being a loner, an analyst needs people skills. This role interacts with people across the company. A good data analyst also has advanced understanding of your business intelligence tool.
It doesn’t matter which role you hire first. What matters is that you plan to fill both so the spectrum of skills is covered.
Look for horses, not unicorns
Since your first hire will carry the program alone for a while, they need to have an understanding of both roles. You’re either looking for an engineer with some analytical skills or an analyst with some engineering skills. They will design and build your data infrastructure and they will develop processes to deliver on data requests.
Just because they understand both roles, don’t think they can carry the program forever. You should make your second hire as soon as possible after the first. Each role needs time and space to focus on what they do best.
It is possible to find a unicorn candidate who has all the skills you need, but those professionals are
- a) Rare,
- b) Expensive, and
- c) Not interested in being a one-man show.
Any data professional with enough experience to cover both roles is also experienced enough to know they don’t want to. A one-person data team spends all their time putting out fires and responding to service tickets. The job leaves no time for the analysis and insights you hired them for in the first place.
If you don’t have useful reports and insights within six months of starting your data team, something is wrong. Either the team, the tools, or the infrastructure isn’t working correctly. Reassess. Make sure your expectations are realistic – that analysts are charged with analyzing, engineers are charged with engineering, and no one is expected to work magic.
Lauren Balik is the founder and CEO of Upright Analytics, a full-stack data engineering and analytics firm. She makes complex data topics easy to understand and helps small and medium businesses love their data. Reach her at lauren@uprightanalytics.com.
