Last week, we had the honour of having Lauren Yu, Data Scientist at Facebook host a session on Product Manager/Data Science Collaboration. A small group of product managers and data science analysts with experience from companies such as Food Panda and Grab came together over zoom to share their experience and thoughts on the collaboration between PMs and Data Science. Data Science often means different things to different people, and job descriptions for a data science role can differ greatly from company to company. It was an interesting exchange as the group came together to share their personal stories of how PMs and Data Science collaborate, with an array of experiences across company size and sectors.
We’ve distilled a few key takeaways here for those who missed the session, and there was so much to share that we will likely be holding a second session to continue the discussion - stay tuned for more!
In most firms, the PMs are the owners of the product and act like a mini CEO for that product. They define product direction and make product decisions. Whenever available, such decisions are made based on data, which may come from the Data Science team. In some companies, the data science team is responsible for building basic dashboards for information needs, and in others, the data science team builds predictive models. In yet some other companies, the teams are segregated into different business teams that consist equally of engineers, product people, and data science.
While PMs tend to be the ones that are driving product direction, Data Science can also do the same, especially when they have made observations backed by data. These observations can result in killing off features/products, as well as building entirely new products. For example, when Data Science notices that users are using the product for one particular reason over others, that can lead to the spinoff and creation of an entirely separate product catering to that specific use case.
PMs will always need to know the impact of product improvements and upgrades, and that is where data science comes in handy to run experiments to test and quantify the impact of product updates. In one story shared, the PM was trying to determine the collective impact of 5 unique, simultaneous changes to the product, and needed to find a way to help quantify it. The data science team then suggested running an experiment where they withheld the 5 updates from a small random pool of users, and then ran the metrics comparing users who enjoyed the updates versus users who did not.
There is often no direct reporting structure between PMs and Data Science from one to the other. Different firms are also structured differently, such that in some, Data Science is responsible for product metrics, whereas in others, PMs are responsible for them. Regardless, both parties are almost always equally invested in the success of the product. While this sometimes results in tension when there are different views, it also results in a lot of collaboration. This is especially so when Data Science, unprompted by PMs, discovers certain behaviours or patterns among users that help PMs redefine the target group and/or refine the marketing message of the product.
Therefore, while different firms might designate different responsibilities to Data Science and the interaction with PMs might differ from firm to firm, the commonality was that Data Science plays a crucial role in product direction too, in tandem with the PMs.
Sign up here to get on the waitlist!