Our customer wanted to explore what insights could be gained from usage reporting from their devices (Web platform, TV and mobile).
Since there was not yet a commercially viable use for the data, this was an exploratory endeavour.
We were tasked with defining what could be measured from the devices (in some of them, creating the instrumentation), processing that information, and exploring the data to find actionable ideas.
As there were many events of interest, and many users using the systems at the same time, we decided to batch the reporting of events to ease the load on the server side.
We did not have control of some of the devices, so we had to process the data we received to make it suitable for further analysis. We had to develop software to do this adaptation.
The data contained for example, when the user did an action. We needed to create a model of how the user behaves: are they still watching? have they left the TV on or the player playing?
To collect and analyze the information, we used the Elastic stack (ELK: ElasticSearch, Logstash, Kibana), running on the Cloud for its ease of setup for temporary infrastructure, price elasticity and redundancy capabilities.
We were able to implement the data gathering and did a thorough analysis of the data.
We found that the results were consistent with third party sources of information that were previously available.
Furthermore, this data was useful to improve user engagement and to better focus future development of the client’s products.