Being a product manager in a data startup offers some interesting challenges. As more and more PMs work for data startups understanding these challenges will help PMs survive the strange world of data.
There are two challenges that product managers in data startups face:
- Translating customer problems into analysis
- Presentation of data to customers
Translation to Analysis
Within a data startup the challenge is not turning customer problems into functionality but customer problems into analysis. As many (most?) Product Managers will not come from a data science background this will require the PM to express the customer problem to the data scientists. The problem will be reining in the data scientist from descending into the beauty of maths and ignoring the problem that needs to be solved. You thought keeping developers focused was difficult, wait until you try and keep data scientists focused on the problem.
I’ve found that the following techniques help:
- Learn enough of the maths to pass the idiot test and at least understand the language being spoken
- Encourage the data scientists to explore but also continually re-frame everything they do to the problem
- Have the data scientists articulate the problem they are solving and present it to the broader company
While not perfect, I’ve found that these three steps go a long way towards addressing the problem of keeping data scientists focused on the customer problem and not the delight of the maths.
Presentation of Data
Once you have the results of the analysis, it needs to be presented. Simply showing tables of data doesn’t cut it. You as a Product Manager need to understand intimately the data that is being presented so that you understand the most important characteristics that need to be displayed.
Nor is relying on info-graphics the way to go. The key is to present the data that highlights the most salient facts. Unfortunately, there is no handbook (yet) on the best way to present information that is understandable in large quantities via the web but also information dense (in the Edward Tufte sense).
I have found that an approach of the “rabbit hole” to work quite well. The “rabbit hole” is where you provide a summary bit of information that the user can drill down into to see more detailed information as and when they see fit. Think how Alice falls into the rabbit hole chasing the white rabbit. Figures 1 to 3 show a Rabbit Hole in action.
The hard part is picking the rabbit hole information. This will depend on the audience and also the data in particular what works as summary information that is also intriguing. You want users to explore the data and finding that hook information is often very difficult.
Data startups and Product Management within them brings its own issues. As time goes on we will develop a set of techniques and practices that allow product managers to overcome the challenges of working with data as the product.