Product derived data as a tool for product managers

Making decision without full knowledge of all facts and circumstances affecting possible outputs is part of everyday life. In certain areas getting full picture is too costly or even impossible. There is though lots of data available around us that can be used, in relatively inexpensive way, to guide us with decision making.

One area where different kinds of data is needed on daily basis is product management. For both product development and marketing, knowing how product is received by markets and how it performs is crucial to optimally direct maintenance and development efforts and how to decide on marketing mix.

Among many different sources of the data that can be used by product managers important, but often neglected, is the product itself. Gathering data from product controllers and/or external sensor often brings extra dimensions to our knowledge of how the product is being used, how it performs and how it interacts with environment.

If we collect data from enough products that are distributed among users in representative way, we can picture differences based on geography, customer segment, ambient data and many other factors important in specific situations.

Of course, just getting large dataset of numbers is not going to help product manager in any way. The key is to have processed results and visualizations showing patterns and distributions rather than raw data. Presentation of data should be flexible enough to answer questions like how big percent of users from region A uses function B? Or what are coldest environment product is successfully used? Does ambient temperature correlate with MTBF of the product? Efforts should also be put into finding phenomena that were unknown without looking at the data.

Important area of information may also be codes or values related to abnormal states. Usually, information about problems reaches, event most attentive product manager, after long delay, first travelling from customer through dealer, claims department etc. Also, accuracy of claims is mostly very poor – what exactly was the problem? How often it occurred? In what condition or phase of the process? Having first hand info about the problem may considerably help to shorten reaction time to quality issues and to direct resources to sort out problems that are most important rather toward those reported by shouting loudest.

How would that work in practice? Four steps are needed:

·       Secure data acquisition – get local data and send it to for further processing

·       Data processing and storage

·       Analytics & visualization – to look for patterns, abnormalities and explanations

·       Resources to actively work with the data

 It depends on industry but quite often there will be vast amount of data to collect and process therefore there is need of certain expertise to get something out of the data. Also, there will rarely be enough resources to systematically analyze all what was collected therefore automatic algorithms to look for abnormal situations and patterns in the data are required. Domain experts look then for explanation of marked records or regions, not having to spend time on those more obvious results. Statistical and AI type of tools are helpful but require certain knowledge and experience.

Is product derived data replacing any of the tools product manager uses today? Of course not. Profitability, sales follow-up, competition analyses etc. must be daily focus of product manager. But having near real-time access to product derived data may improve decision making and free up precious time from trying to guess how the product performs and how it is used.