SPSS Analytics Partner | Case Study – Si.mobil

Case Study – Si.mobil


The need

To prosper in a crowded market, telecom provider Si.mobil wanted to find new ways to cut costs and boost revenues. It identified customer retention and handset investment as key areas for improvement.


The solution

Si.mobil deployed powerful data modeling software to reveal deep insights into customer behavior – predicting whether customers are likely to churn and which handsets they are likely to choose.


The benefit

Predicting churn helps Si.mobil boost retention by 10 percent, saving EUR1.1 million a year. Predicting customer handset choices enables the company to cut annual hardware investment by EUR1 million.



Predictive analytics cuts churn and reduces hardware investments, saving millions of euros per year


When you are competing in a crowded market, how can you foster loyalty among your existing customers and attract new ones?
By investing in state-of-the-art data modeling software, Slovenian telecommunications provider Si.mobil is now able to predict which handsets customers are likely to choose, and which customers are likely to switch provider.


The new capabilities play a key role in helping Si.mobil differentiate itself with superior customer service, reduce its investment in mobile hardware, and retain customers through proactive, targeted marketing. As a result, the company can both increase its profitability and protect its market share.


Competing in a crowded market

Competition on the Slovenian telecommunications market is fierce, with four mobile operators and three mobile providers contending for the largest share of the nation’s two million consumers. As competition has intensified, the price of telephony services and mobile devices has dropped – and operators’ profit margins have been squeezed.


Operators that can find ways to reduce costs, secure the loyalty of existing customers, and attract new ones stand to gain huge advantage over their competitors.

Full backing from the company’s senior management was a key factor in the success of Si.mobil’s analytics implementation. “Our executives recognized the benefits that the SPSS solution could bring to our business, and placed analytics at the heart of our corporate strategy,” says Elvir Mujkic, Head of Infomanagement at Si.mobil.

Solution components

  • IBM® SPSS® Modeler Gold Server Edition

Andreja Stirn, Director of the Business Intelligence Center at Si.mobil, explains: “To differentiate our company from other providers, we focus on delivering the best customer experience. This requires us not only to help customers choose the handsets and tariffs that best fit their needs, but also foster their long-term satisfaction by providing excellent network coverage and availability. We decided to invest in analytics to help us achieve these twin aims.”


Finding the key in analytics

Having already built a data warehouse to store data from its business systems, Si.mobil decided to purchase IBM® SPSS® Modeler Server, an advanced analytics platform that would help to model and predict future customer behavior.


Elvir Mujkic, Head of Infomanagement at Si.mobil, adds: “When it came to choosing a solution, we selected SPSS, because we felt that it was very user-friendly and offered better value-for-money than competing offerings.


“We had been using the desktop version of SPSS Modeler for a few years for some small projects, but upgrading to the server edition has enabled us to automate the process, increase the number and variety of models we create, and deliver insight on a weekly or even daily basis.”


Reducing customer churn

With the cutting-edge modeling solution in place, Si.mobil turned its attention to reducing customer churn. The company was keen to identify the customers who were most likely to switch to another provider, and target them with specific marketing offers to encourage them to stay with Si.mobil.


The modeling team established 53 key performance indicators that correlate with customers who are likely to switch to another provider – including factors such as where a customer lives, how often they call or text, and whether they have called one of the company’s competitors. The team then built two churn models – one for customers who have monthly contracts and one for customers with pay-as-you-go phones – that categorize customers as high, medium or low risk of churn.


Si.mobil then targeted the highest-risk categories with rewards (ranging from extra minutes to new handsets) to encourage them to stay with Si.mobil. Offering incentives only to customers who are likely to switch helps the company keep costs down and achieve a greater return on its marketing spend.


Andreja Stirn continues: “With pay-as-you-go customers, the key is to predict whether they will top up. Out of the people we targeted with incentives, 5,000 continued to top up and were still active customers by the end of the year. If we extrapolate these results, we could boost revenues by EUR 400,000 per year.

“IBM SPSS helps us provide the level of service our customers demand – while unlocking
multi-million-euro savings for the business.”

— Andreja Stirn, Director of the Business Intelligence Center, Si.mobil

“When it comes to long-term contract customers, what we want to know is whether they will renew their contracts. By offering greater rewards to those at higher risk of changing provider, we have been able to reduce churn by 10 percent. This translates to a revenue increase of EUR1.1 million per year – a significant contribution to our EUR200 million annual revenues.”


Predicting which phones customers will purchase

Next, the company shifted its focus towards making smarter procurement decisions.


While contract customers generally pay for new phones over the course of a 12- or 24-month contract, operators need to buy all of their handsets up front. This investment represents up to 25 percent of Si.mobil’s total costs – and creates a significant risk if customers fail to pay their bills. From an operator’s point of view, therefore, it makes
sense to encourage customers to choose less expensive handsets, to minimize both the investment and the risk.


Si.mobil set out to predict which handset each customer was likely to purchase, looking at factors such as how much they are likely to spend, which handset manufacturer they would select, which operating system they would choose, whether they would purchase the handset outright or via a monthly payment on their bill, and how often they call, text or use mobile data.


By examining these factors, Si.mobil can establish a shortlist of phones that each customer is likely to purchase – and offer this shortlist to the customer, rather than the entire range.


When Si.mobil tested the effectiveness of its model, it found that older customers who received personalized offers were 10 to 15 percent more likely to respond, and approximately 50 percent of them picked one of the handsets we offered on the shortlist.


Andreja Stirn comments: “Presenting customers with the right options makes them much more likely to choose an appropriate phone for their lifestyle. Not everyone wants or needs the latest iPhone! And by reducing the need to invest in so many high-end handsets, we expect to save up to EUR1 million per year. Considering the size of our market, this is a remarkable result.”


Focusing on the customer experience

As Si.mobil begins harvesting the fruits of its investment in analytics, it is now planning to use the SPSS solution to tackle challenges in other areas of its business, such as the quality and coverage of its network.

Andreja Stirn explains: “Call quality and network coverage are hugely influential in securing and maintaining the satisfaction of our customers. SPSS can help us improve the quality and availability of our network by revealing which customers are experiencing problems, whether certain devices receive worse reception than others, and whether coverage is poor in certain regions.”


Si.mobil is planning to combine data about the state of the network with faults reported by customers or discovered through behavior analysis. This will enable it to identify the errors that cause real problems for users. The SPSS platform is a key enabler for this type of large-scale analysis, which will require the company to match seven million daily call records with hundreds of millions of network events.


“With our findings from IBM SPSS, we will be able to provide vastly improved customer service – and unlock million-euro savings at the same time,” concludes Andreja Stirn. “We are confident that the solution will provide a key source of competitive advantage in the months and years to come.”

© Copyright IBM Corporation 2014