I took on very different roles and responsibilities while doing my corporate data science jobs. Not only did they give me a lot of exposure on data science, but they also taught me how to manage my managers.
I remember one of the roles very vividly. Reported to a person who had never done a data science project before or led a data science team. He was a good person and a good manager in other settings, but in the wrong place to lead a data science team (at least temporarily). Most of his experience prior to taking over this role was in Sales. Some of the alternatives for managing the people who worked for him in the past will not work in this configuration.
As an example, we used to have 2 team meetings in a week, where each of us updated the rest of the team on what we are working on and what is the plan for the rest of the week. Beyond these, there were individual updates and updates associated with the project. Most of us did not understand the goals of those weekly assignments..
The situation I have described is definitely not unique. This would be happening in all institutions trying to determine a data science team or trying to transfer some of their best workers to other roles to lead data science units. Therefore, I thought I'd share some tips for people making these transitions. These tips should help you become a better data science manager, if you haven't been in a data science setup before.
Be part of the full life cycle of the analysis project / data science:
Nothing beats this advice. No matter what research you can get from the experience you get from being part of the team doing the project. You will understand why marketing analytics to a potential can be difficult at times and it may take months before the client gives you a fair chance. It will also help you understand why data cleansing can take, what seems like an eternity from the outside. At the same time, implementing an analytics solution can give you its own part of learning: What are the hindrances you can find? Why do you need to be paranoid about successful implementation??
If there is a tip that you want to get from this post, just take this. You can't lead a data science team effectively until time has passed (preferably practical) working on a project yourself..
Understand the data science landscape:
Although the first tip helps you dig deeper into the subject, you must also understand the extent of the topic. That's what differentiates a good manager from a brilliant analyst.. as manager, you need to understand which tool and solution is best for what kind of problem.
Does data demand a big data solution? Or does a traditional data science method work? Do you continue to automate reports in Excel or switch to tools like QlikView or Tableau? These are some of the questions you will face as a manager and the decisions you make will impact your team members and how they spend their time.
This post can be a good starting point.
Become awesome in structured thinking:
This is almost a fact: you can't be a good analytics manager if you're not good at structured thinking. As an analyst, expected to put structure to unstructured problems. as manager, expected to excel in putting structure in place, in minutes. You would enter meetings that would lack structure and would only benefit from them, if you have the ability to put a structure to the discussion.
These posts can help you improve structured thinking: The art of structured thinking, Tools to Drive Structured Thinking
Improve your storytelling skills (supported by data):
As a data science manager, expected to explain data-driven stories. The basic expectation is to be good at communicating your thoughts.. A good manager must be able to visualize data effectively and present it in a way that tells a coherent story. Here are a couple of examples to get your brain thinking:
- If you have to understand the regional distributions of your product, What is the best way to understand it? Tabulate penetration by regions? By bunches? Or just draw a heat map superimposed on the geographic map?
- What is the best way to show how sales have changed in 15 groups compared to last year?
The better you become familiar with storytelling, the better it will be for your team to market the solutions.
Define your own learning plan and agenda and share them with your team:
If you are assuming a data science administrator role, has a lot to learn in the next few days. The best way to do this is to create a learning plan and share it with your team.. This will not only help them understand what you already know and what you don't, but also communicate that you are willing to learn the subject.
You can also ask team members to create their learning plan and share it with the group and have weekly knowledge sharing sessions with teams to share their learning. / experience.
Final notes:
What do you think of these tips? Are you being prepared to push through the management of a team of analysts or data scientists?? If you have more advice from your experience, feel free to share them through the comments below.
Image source: iNostix blog
