Case study: Validating big data use cases
Case study: Validating big data use cases
Who? Robi Axiata Limited, second-largest mobile operator in Bangladesh
What?
Wanted to determine feasibility of using big data analytics to reduce churn, increase revenue and improve customer experience
How?
Used TM Forum’s Big Data Analytics Guidebook and participated in Catalyst to evaluate use cases
Results?
Robi is ready to move forward with implementing a big data analytics platform Like most network operators, Robi Axiata Limited, a joint venture of Axiata Group and NTT DoCoMo and the second-largest mobile operator in Bangladesh, is interested in figuring out how it can use big data analytics to improve customer experience and increase revenue. But rather than jump headlong into purchasing a big data analytics platform, the company decided to analyze the feasibility first by consulting global standards and examples on how a big data project should be approached. TM Forum’s Big Data Analytics Guidebook (GB979) is one of set of examples the company used to do a proof of concept.
“We started our project in June 2013 to understand what big data is, what the current trends are, what the business challenges are that operators are trying to solve and what other industries are doing with big data,” says Ahmed Saady Yaamin, Vice President, Business IT, Robi Axiata. “It was kind of an R&D initial feasibility study.”
While Yaamin says he is not able to share specific results of the proof of concept, they were “better than our initial expectation” and included a reduction in churn and a better uptake of services. This led the company to conclude that it is worthwhile deploying big data analytics on a wider scale. Robi is now moving forward with implementing a big data platform.
“The experience gave us confidence that, yes, we can do it,” Yaamin says. “But it also taught us that you need to start small and you need to be prepared to refine the use case model quickly. It’s very important to make sure that whatever insights you get you implement into actions. If the [internal] users don’t see any value, they won’t cooperate. Therefore, it is very important to implement a closed-loop implementation model during the initial phase.”
Setting priorities
The Robi team realized that the traditional way of documenting analytics use cases wouldn’t help in a big data project, so the team decided to come up with a format that was more robust and comprehensive. The team turned to TM Forum’s Big Data Analytics Guidebook. While the current version of the guidebook offers 48 use cases, the version Robi used offered 30, and the first step for Robi was prioritizing the use cases based on the company’s strategic goals.
“Because the use cases were coming from a well-known organization like TM Forum, we knew they had been vetted,” Yaamin says. “We just needed to customize them for our situation. In addition to that we decided that we should prioritize these use cases based on business needs.”
Yaamin’s team plotted potential use cases in a matrix with the X axis representing the impact of the use case on the business (for example, customer satisfaction has a high impact) and the Y axis showing the probability of implementing a particular use case. Use cases with higher probability and impact get priority, Yaamin says.
“If you have a use case that uses data about voice-dominant subscribers, we have that data,” he explains. “But if you want to do one on social media, it would have a lower probability of implementation because we don’t have the right data in hand.”
Yaamin’s team prioritized a total of about 45 to 50 use cases covering all functions such as sales, marketing, network analytics and business forecasting. About 20 percent were deemed high-probability and high-impact.
“After we came up with the prioritized requirements, we wanted to implement two use cases on our own for hands-on experience,” Yaamin says. “We’ve done that, and in parallel we are going through the procurement process to find a vendor who can deliver on four to six uses cases in a phased deployment.”
Meeting customers’ needs
The first use case Robi implemented sought to identify customers prone to using more data and suggest the best data pack for them, based on their usage behavior. This involved building a ‘data customer journey map’ to find out why customers are churning and how to offer them the best data pack to suit their needs, according to Yaamin.
“Initially we used all the typical data an operator has access to, and on top of it we gathered some information from social media by asking customers questions,” he says. The company analyzed the data it collected to figure out which customers might be interested in buying high-end data packages. “Then we monitored usage and uptake and tried to fine-tune it,” Yaamin adds. “If we delivered a high-end data pack to a customer but after a week of watching his usage realized he was truly a medium user, we fine-tuned the package based on his needs.”
The second use case involved working out how to change Robi’s call center from a cost center into a profit center. “We tried to change it to a profit center by prioritizing calls that can generate sales leads.” Yaamin says. “A subscriber calling to understand a current promotion is more likely to generate revenue than someone calling to issue a complaint.”
All complaints are not equal
The team recognized that dealing with some complaints was more important than other. So they went further and tried to categorize complaints based on the customer’s propensity to churn following each type of complaint (for example, issues with call quality, network reliability or broadband service). The company analyzed six months’ worth of data to figure out which types of complaints typically lead to churn.
“A customer who is complaining with problem that has higher churn propensity should be prioritized and handled a different way,” Yaamin explains. “If we can reduce the churn of those customers, eventually that gives us more revenue.”
Catalyst for change
Robi also participated in a Catalyst project in Nice, Big data, big profits: Harnessing the power of analytics, in order to learn more about big data analytics.
“After we got the guidebook we studied it but it was one-way communication – when we had question there was no one to answer,” Yaamin explains. “The Catalyst gave us an opportunity to interact with other operators and solution providers.”
Using the guidebook and participating in the Catalyst project convinced Yaamin and his team that implementing big data analytics will be worthwhile, but they also learned that deploying the technology will not be the most significant challenge.
“Until you get buy-in from management and the business users of the project, it’s hard to make it successful,” Yaamin says. “It’s only successful when users are finding value in it and are excited to use it.”
Yaamin and his team are working now to evangelize big data analytics internally.
“When you ask people to implement new insights and analytics, it’s a big threat to their comfort zone,” he says. “We are trying to make them understand that we are here for their benefit by explaining that if they put analytics into daily action, they can increase revenue and solve business challenges.”