Kyvos Insights > Case Studies > Viewer Analytics: Interactive BI on 168 Billion Subscriber Interactions for Personalized Content

Viewer Analytics: Interactive BI on 168 Billion Subscriber Interactions for Personalized Content

Company Size

1,000+

Product

Kyvos

Tech Stack

Hadoop
Tableau
Qlik

Implementation Scale

Enterprise-wide Deployment

Impact Metrics

Customer Satisfaction
Productivity Improvements

Technology Category

Analytics & Modeling - Big Data Analytics
Analytics & Modeling - Real Time Analytics

Applicable Industries

Telecommunications

Applicable Functions

Sales & Marketing

Services

Data Science Services

About The Customer

The customer is one of the world’s largest telecommunications companies. As part of an initiative to improve experiences and deliver more value to their viewers, they wanted to analyze viewership data from millions of television subscribers. They aimed to understand their data across mediums, viewers, geographies equipment, and use those insights to power their marketing decisions and deliver personalized content to their subscribers. The company had access to a massive amount of session data from live TV viewing, DVR, video-on-demand, pay-per-view, set-top box usage, and other streaming devices.

The Challenge

The telecommunications company had access to a massive amount of session data from live TV viewing, DVR, video-on-demand, pay-per-view, set-top box usage, and other streaming devices. However, it was difficult to connect and get useful insights from this data. The existing infrastructure worked well for small datasets, but as data volumes grew, they started facing challenges in processing and deciphering it. The company was unable to leverage the massive volumes of viewership data from millions of subscribers stored on Hadoop to improve the viewer experience and gain a competitive advantage. New or ad hoc queries took hours or even days to return, making them unusable for decision-making. Slow responses to queries made it impossible to analyze data over extended periods to understand trends or recognize patterns.

The Solution

The company built a BI acceleration layer on its data platform using Kyvos to enable instant access to all the data on Hadoop. Smart OLAP™ technology allowed the pre-aggregation of data into cubes across multiple dimensions. As all combinations were processed in advance, they provided instant responses to all Hadoop queries. The same queries that took hours could now be returned in seconds. Kyvos integrated with their existing BI tools, Tableau and Qlik, providing a familiar analytical environment for analysts and business users. It allowed them to analyze data within Hadoop interactively, across many dimensions, with instant response times. This helped them achieve high performance and solve problems with data at scale.

Operational Impact

The company was able to get instant insights into millions of subscribers, improving experiences dramatically and delivering personalized content to viewers.
Quick answers to business questions provided the agility needed to respond to network, content, customer, and company problems.
The company was able to use any BI tool as a native Hadoop BI tool.
Timely and accurate analysis on 14 months of data as compared to 1 month earlier.
Exceptionally high performance with queries returning in seconds as compared to earlier response times of 5-10 minutes.

Quantitative Benefit

Built a 70 terabyte cube with 14 months of data, 38 dimensions, and 168 billion fact rows.
Query performance improved drastically. While traditional OLAP tools took 5-10 minutes to return a query on one month of data, Kyvos returned queries on 14 months within 4 seconds and often in less than a second.
Cube refreshes were fast and automated. With around 250 million rows added daily, the cube's daily refresh time was less than 20 minutes.