Big Data Analytics and Visualization

The market of digital solutions and services for improving customer experience is crowded and diverse. Therefore, it’s hard to present a unique product in this competitive space. Hard, but not impossible.

Feed.fm created an innovative solution for business to augment digital experiences with motivational and engaging music accompaniment tailored to audience’s preferences and situational needs. The company allows the network of clients to integrate their apps and web systems with Feed.fm platform and enhance people’s digital experiences with authentic media background.

Our work with Feed.fm could be one of the best data visualization examples which brought viable results. Feed.fm has been collecting datum since 2015. In fact, big data became one of the core platform’s influencers that make music enhancement truly user-oriented and impactful, conversion and business-wise. In this respect, both legacy and real-time data need representation and interpretation. That’s how Digiteum was assigned to create Feed.fm big data analytics and visualization and help the company and its clients turn raw playback data into valuable insights. Later, this case became a good big data visualization example to show how interpretation of data can help a business or organization grow.

At the first stage, Digiteum used the combination of Elastic Stack solutions (Elasticsearch, Logstash and data visualization application Kibana) to enable platform’s big data initial collection, systematization, analysis and visualization for both Feed.fm team and the company’s clients.

On the one side, the information from individual application media playback – sessions length, number of unique listeners, audience behavior and preferences, conversion rate, etc. – comes to the client’s individual login-protected admin dashboard. This data represented in visual charts and comparison-based graphs provides real-time performance reporting and allows the users of Feed.fm platform monitor and control the efficiency of their applications. You can see playback data visualization example in the image above.

On the other side, big data combined from all clients and applications comes to Feed.fm’s internal dashboard. This feed allows editorial teams to compile custom track lists relevant to clients’ needs and their customers’ expectations, continuously improve the technology behind the platform and eliminate risks, and in general, analyze and understand how media augmentation affects customer engagement, loyalty and conversion.

The next stage of the project required advanced data analytics and management tools and, as a result, enabled more profound, insightful work on the platform’s data.

At this step, Digiteum team first takes data and statistics from Elastic Stack, processes this data using Apache Spark analytics engine and builds various aggregations – data sets associated with one platform user that illustrate certain performance patterns in given term – day, month, etc. For example, these aggregations can show how music playback activity influences user engagement or how much time active users who play music spend in the app compared to less active users. This also is an example of data analytics which applies advanced algorithms. In order to determine patterns and build aggregations, the team relies on mathematical statistics (one of the methods used in Machine Learning to find dependencies and make predictions).

Then, easy-to-read and understand aggregations with data insights are stored in Apache Hadoop HBase. This storage solution is perfect for time series data produced by Feed.fm for many reasons, including data optimization capabilities and SQL integration. Visualization of these insights will take place at the next stage of the project and will make another cool data visualization examples that illustrate how business can benefit from big data.

Advanced analytics, pattern recognition and soon visualization of more detailed and complex data insights have significant value for both Feed.fm, the platform’s clients and millions of end-users. On one hand, it allows Feed.fm team to quickly interpret reaction to changes in platform performance and adjust improvement efforts for great results. On the other hand, the clients of Feed.fm get deeper insights on customer engagement that allow them to make relevant data-driven decisions and choose the approach to web or app optimization, conversion improvement and profit maximization.