Product Managers are supposed to be metrics driven and are used to extracting and analysing data to understand how the product has been performing – for the customers/users and for the business. This is usually done as an operational exercise on a day to day basis or sometimes to figure out the cause of something unusual happening on the product. This is a typical usage of dash-boarding and business intelligence; perfectly fine if you do not (or cannot) collect data beyond a few key metrics.
Today, most enterprise and consumer products collect and store enormous amount of data on the usage of the products. The usefulness of this data goes much beyond dashboards and business intelligence visualisations. Product Managers – who are serious about their products should take the help of Data Scientists – specialists who know to combine statistics, data mining, analytics to extract knowledge out of data. Usually some Product Managers find it difficult to find out when and how to collaborate with a Data Scientist. My experience with a few products which collect “huge” amount of data has been that Data Scientist can help in two situations:
a) When you know what you want out of your data and b) When you want to explore about what to do with your data
When you know what you want out of your data: This is a part of product development or operations. Products need slicing and dicing of data to understand user behaviour. Without a Data Scientist one would end-up looking at the major metrics and struggle to understand several underlying behaviours. You need a data scientist to figure out reasons for the symptoms and institutionalise a framework such that it is visualised and understood. There are also cases where products need to separate the wheat from the chaff and the accuracy of achieving that measures the success of the product;Data Scientists help you in working with the development team to ensure that it is done at scale with minimum errors. In hypothesis based experiments, the expert helps you to make sure that you reject or accept the hypothesis with data that is statistically significant.
An example of this type of project would be when you want to extract FAQs from mailing lists – where you need to device a model to efficiently generate the FAQs from a pile of user generated content.
When you want to explore about what to do with your data: Customer or user interviews usually bring back a truck load of anecdotes, ideas, issues and requests. Similarly market and competition throws at you plenty of data points. They say, anecdotes are plural of data, but such things are easier to say. A Data Scientist can help you find the hidden meaning in the data which can help you create a product roadmap. This is more of an exploratory exercise where you need to work more closely with the Data Scientist than in the previous case.
An example of this type of project would be when anecdotal evidence suggests that high bounce rate of users for your web application is due to bad first time user experience; before you embark on improving the experience, data might be able to tell you if such a pattern really exists in at least some form.
Finally, do not forget that Product Manager is the custodian of the product. Data Scientists are specialists who can help you achieving the product goals just like how a Software Engineer or an Architect would.
Picture courtesy: Bragg