Not a data analyst, but need to be data-driven?
This is a very common question I’ve encountered from Product managers who don’t have a data background, but want to make data-driven decisions. The most common answers I’ve heard to this question is — well, you’ve got to learn SQL — and that’s not the full truth, in my opinion.
Being data-driven is largely a mindset. Not having the ability to write a SQL query shouldn’t hold you back from making data-driven decisions.
While knowing SQL and worked on R/Python/SAS in the past has it’s advantages, not knowing them doesn’t mean you can’t be data-driven in the product decisions you make. Being data-driven is a way of being, an obsessiveness with curiosity that takes time to nurture, a way of thinking differently.
PMs who are hands-on with their data typically are very involved in the process of how that data is being generated, where it’s being generated, how it gets transformed and how they want to consume it. In fact, most of us who use data to make product decisions need reliable data to do so, and are immensely invested in the quality of the data we are consuming.
This means, many a time, we work back from the data we need and set requirements around how, when and where the data needs to be created. Most of us likely create our own dashboards and learn how to, even if we don’t need to. We work with anything that will help us run quick analyses, using whatever tools we can, sometimes Excel becomes our best friend.
Making data-driven decisions everyday has a few key components:
Start with an objective: Data can take you down many rabbit holes. Know what you want to answer with the data, and stick to the questions. You will most likely find many interesting nuggets of information as you explore the data, but this way, you’ll stick to the primary trail instead of deviating into forks that are ultimately cul-de-sacs.
Storyboard your data: There is a scientific, methodical way to breaking down data. Just like we storyboard the user flows, data needs to be storyboarded. You will find information and slowly put together the right order of findings, proving causality which ultimately brings you to a comprehensive answer to the question you started out with.
Keep asking why: This is very important while working with data. Data has a way of telling a story. But the story is a treasure hunt. You can start with one data point, and be satisfied that you’ve got the answer you’re looking for. However, the truth, more likely, is that you’ve just found the tip of the iceberg and the more you follow the trail, the closer to the truth you will get.
Second guess yourself: Check, double check if your numbers make sense. Do they add up? Are you reading them right? Most of the time, second guessing isn’t an asset to product managers, but with data, it often is.
Double confirmation: Can you verify the story your data tells you with survey data or A/B tests? Data-driven decisions are powerful when combined with these tools, so you can verify that you aren’t relying on incorrect or incomplete data and flying blind without knowing it.
Data has a way of telling a story of it’s own. Remember, you can always get data to tell you the story you want to hear. Making sure you are getting the real story is your job.
Of course, the reason why everyone answers with “you have to learn SQL”, or “you’ve got to have experience with a BI tool” because knowing your way around these tools puts you in more control of the narrative. Not only that, knowing these basic tools can help you adapt to different languages and tools that companies use without sweating it. If you know SQL, you can figure out how to read Splunk logs fairly easily. As a PM, that gives you a lot of horse power and self sufficiency to run analyses on your own schedule, continuously, without having to depend on anyone else. They also make you more aware of how your data flows and help you manage and maintain the quality of the data you use. And finally, if you’re in a data immature environment, it gives you the ability to pull data across multiple warehouses and find the data you need.
If you are building products that use machine learning, you need advanced understanding of what data is being used by your product and how that impacts the customer experience. Even if you can’t model yourself (and you shouldn’t have to beyond Excel regressions and forecasts), you should develop a basic understanding of statistics if you want to be an effective PM for ML products.
To that effect, I’ve listed out a few (and definitely not comprehensive) resources that might help you get started with data as a PM.
Resources:
Here’s a list of resources for PM’s wanting to learn to work with data:
2. https://productschool.com/blog/product-management-2/machine-learning-google-pm/
3. Data science with R: https://www.kdnuggets.com/2013/02/free-e-book-on-data-science-with-r.html
4. One of my favorites — Python for Data analysis: https://www.oreilly.com/library/view/python-for-data/9781449323592/
5. Data science for business: https://data-science-for-biz.com/
6. On testing: The startup way https://www.amazon.ca/Startup-Way-Companies-Entrepreneurial-Management/dp/1101903201
7. Courses: https://www.coursera.org/learn/data-science-for-business-innovation
8. ML course: https://www.coursera.org/specializations/machine-learning#courses
