Telecoms Understand Customers with Customer Behavior Analytics in SQream DB
Calculate more KPIs, faster with SQream DB
One of Asia's leading Tier-1 mobile operator deployed SQream DB to unlock more insights from their collected data. This led to increased revenues, by implementing an advanced location based analytics solution for targeted billboard campaigns, with optimized ad targeting and near real-time interactions.
Overloaded, distributed system before SQream DB
Before deploying SQream DB, the operator had an infrastructure consisting of 40 active Greenplum nodes. The lengthy ETL process limited daily reporting to a short window, reducing relevance of reports due to the aged data in the system.
Simplified analytics solution with SQream DB
The operator chose to deploy SQream DB for scale, speed and simplicity afforded by its market-leading GPU-based design. SQream DB provides a high-performing, rapid, end-to-end analytics solution enabling the marketing business unit to get deeper near real-time insights about its customers and their behavior and by such, deliver them with more targeted, customized ads.
The new SQream DB system avoids the unnecessary pre-aggregation steps, and optimizes the ETL process, simplifies the complexity of the database infrastructure and allows data to flow constantly between the source systems and SQream DB.
Offering better products to customers with customer behavior analysis
Implementing customer behavior analytics with a GPU database - the operator was able to see, for the first time, the correlation between churn rate, user retention and number of support calls at the same time.By combining data from a variety of Hadoop and other relational databases like Oracle, the operator was able to select key customer segments, and with a click of a button identify the level of service they were getting.
- • What are the most popular offerings? What network equipment was used?
- • Were 4G users connecting to slower 2G or 3G base stations?
- • How was the network experience in malls over the weekend?
- • Where did customers use Instagram most?
- • What malls are most popular, broken down by age group and gender?
These are some initial KPIs that the customer analyzed. They built specific queries and Tableau dashboards for these.
For the future expansion, the plan is to carry out complete funnel analysis in the marketing team, to identify how effective targeted ad-campaigns are:
- • How many customers did a specific billboard reach?
- • Are customers who had physical brick-and-mortar stores in their vicinity more likely to stay on the network?
- • Which valuable Diamond-level customers came from specific ad-campaigns?
“I can click on things, change parameters, look at different geographies... I filter differently, and therefore save a lot of time on insights for our business teams.”
Time-to-market for reports reduced by 18x
SQream DB ingests tens of billions of records per day, and can query both fresh and historical data at the same time, without lengthy pre-aggregation steps.
TCO reduced by 90%
By using a repurposed 2U server with a single NVIDIA Tesla card for handling 40TB of data, the operator is now able to avoid expanding the legacy MPP system. With a leaner IT operation, there are now tangible savings on expenses while delivering the business new analytics capabilities.
TCO was therefore reduced by 90% compared to the legacy MPP system.
Enthusiasm about running queries
When the operator's users first started using SQream DB, they immediately became enthusiastic about it. Satisfactions levels are high, because SQream DB is so easy to use, and is much more flexible. Without the constant need for cube-generation and pre-aggregation steps, business users can change their queries many times per day, without having to plan ahead.