It's no surprise that we live in a fast-paced world. There is tremendous focus on always-on mobile connectivity, and access to information when and where we want it, both in the consumer and business worlds. Telecom operators are seeing this new normal firsthand, providing data networks and technology convergence to bring information and access to their customers in the moment. As a result, it's become critical for operators to have the same type of access and interaction with their customer data and analytics. The current competitive landscape demands it.
At the same time, as markets get increasingly saturated, it's difficult for operators to grow by gaining new, first-time mobile users. To increase acquisition numbers, telcos regularly target their competitors' customers with low-priced acquisition offers, decreasing their share of high-value acquisitions. It's become a volume game when it comes to customers. A game that's not sustainable.
When it comes to retention, it's more difficult than ever to defend price premiums based on good service because of the narrowing service-quality gap among operators. What's more, many operators are forced to offer discounted retention incentives, further decreasing the average revenue of existing customers. Add to that the fact that it's easier than ever for a consumer to switch providers, especially with social media influencing decisions by broadcasting true customer experiences and product reviews. It's a recipe for customer migration and lower revenues.
Product upsell and cross-sell is also a challenge, since customers clearly want to know the benefits of the product for which they pay extra money. Meanwhile, voice revenue continues to decrease and data revenue growth becomes more important with VoIP, Internet, and mobile data gaining prominence.
With all these new challenges, operators find themselves in a tough situation: It's becoming increasingly difficult to continue sustainable revenue growth. Many agree that using customer insight correctly is essential to overcoming these challenges. But are traditional analytical models enough? No. If operators are changing the way they do business, this shift must apply to analytics, as well.
Current analytical modeling uses and drawbacks
Analytical modeling is widely used in the telecom industry to tackle difficult tasks such as churn prevention, increasing average revenue per user (ARPU) through cross/upsell, optimizing the customer experience, creating targeted campaigns by understanding customers' needs, and fulfilling those needs with the most relevant products and services—all to increase ROI while maintaining customer satisfaction. In addition, analytics helps to identify and segment groups of customers with similar characteristics, as well as provide marketers the insight required to craft marketing programs with differentiated offers for each of those groups. Analytics has become the backbone of CRM strategy through its ability to obtain valuable information from customer interactions and behavior, then convert it into customer loyalty and revenue uplift.
Analytical models can be classified into two methodologies: descriptive modeling and predictive modeling. Descriptive models are basically segmentation models such as value, behavior, and needs that are developed from a high-level, strategic standpoint. Or, they're more detailed or focused segmentation approaches, such as RFM models (recency-frequency-monetary value), used to solve tactical issues. A commonly used example of strategic segmentation is value-based segmentation, where customers are grouped according to the revenue and profit they bring to the operator, then placed in a customer management and customer service tier according to these segments.
Predictive models are developed by applying advanced data mining techniques such as decision trees, regression, and neural networks to estimate what is likely to happen in the near future based on realized past events. The biggest area of use is churn prevention, which predicts which customers are likely to leave the operator by analyzing the difference between those who have left previously and those who stayed.
Pre-churn behavior analysis looks at a considerable amount of customer information, including call behavior, demographics, payment/recharge properties, customer contact details, and campaign responses. Customers with a high probability of leaving the operator within a set period of time (e.g., two months) are identified and proactive retention activities can be applied. Depending on the maturity level of the market, separate models can be customized according to different customer segments, which allows operators to develop an understanding of what is causing churn in each segment. For instance, the concept of churn should be maintained differently in the markets where mobile number portability is available and cus-tomers can change their operator without changing their mobile number than in markets where number portability is not available.
Another example of predictive modeling is propensity models, which are statistical methods used to identify the best-fit customers for a specific product or service in terms of likelihood to buy. Propensity models allow operators to increase the effectiveness of their campaigns by increasing the response rates and minimizing annoying their customers with irrelevant offers.
These analytics programs, accompanied by an effective customer contact strategy, are a valuable way to improve campaign efficiencies and communicate the right message to the right customers. However, they typically run only monthly, often using historical information from six to 12 months prior. They provide a good general picture about the customer (e.g., type of customer), but lack a snapshot of the most updated customer status. In addition, refreshing the models just monthly some- times misses the fast-changing dynamics of the telecom industry. For instance, a customer can be affected instantly by an acquisition offer from a competitor operator. Customers tend to make quick decisions, which are reflected in their daily transactions.
The dynamic nature of today's telecom customers requires an equally dynamic use of analytics to understand customers and make decisions. It's become imperative to monitor the instant changes in customers' behaviors and match them with the most relevant offer as soon as the customer needs it. This is achieved with a sophisticated blend of analytics and business sense.
The new frontier in dynamic analytics
Two new dynamic analytical capabilities—transactional behavioral analysis and capturing data potential—can give operators real-time insight about their customer activity. Tracking valuable customer information, such as recharge balance or daily usage, and understanding customers' real-time activity present huge potential revenue generation opportunities. These dynamic analytics enhance traditional data mining models by incorporating up-to-date information and exact offers that will fulfil customers' needs in near real time.
The first dynamic analytical capability, transactional behavioral analysis, groups customers by behavior patterns, then tracks their characteristics and behaviors. This information is used to target customers with appropriate offers and interactions in an effort to keep and grow their business.
Take the "occasional user," for example. This customer group spends a small amount of money quickly (pres-umably buying mobile minutes), then stays inactive for long periods of time (see Figure 1). Knowing that this customer will purchase mobile minutes periodically (e.g., only when he needs to use his phone), it is fruitless to bother him with regular offers. Instead, it's better to send him small upsell offers while he's actively using his phone. Transactional behavioral analysis can help to identify similar profiles by deeply analyzing the trends within customer information, such as recharge, payment, and usage. The biggest advantage of the methodology is the ability to determine the exact time of action and the details of the offer to be given based on moments of change.
Or consider a group of prepaid customers who consume their credits within 20 to 30 days and then purchase more (called topping up). Pattern analysis revealed that these customers changed their consumption pattern and began to consume their credit in less than 10 days (see Figure 2). However, they continue to top-up with the same frequency. In addition, after consuming all their credit in 10 days there is also an increase in their incoming calls, which is a sign of "bipping." Bipping is when a customer runs out of minutes and requests the other person to call him. There is big revenue potential in this case. A top-up campaign giving a bonus within a limited time can move the customers' top-up time to be more in line with their behavior, thus increasing revenue.
The key success factor in this type of approach is the effectiveness of the campaign execution and response. With many different customer profiles and scenarios, the measurement of the campaign responses and ROI allows operators to increase their use of successful campaigns while trying to improve the less effective ones. The resulting campaigns will produce strong business performance in terms of revenue uplift and churn reduction over the long term.
The second dynamic analytical approach is called capturing data potential (see Figure 3). With this methodology, companies use real-time data to estimate customers' potential interest in a product or service. We recommend starting with mobile data services, one of the biggest revenue streams for mobile telcos now and in the near future, based on ARPU levels. It's possible to increase data revenues by upgrading existing customers' data usage, as well as realizing the untapped potential in non-users to increase data services penetration.
For existing customers, how many megabytes of data the customer has used is not a sole indicator of potential. Therefore, it's necessary to understand the motivation behind mobile data services usage (see Figure 4). The underlying motivation can be business use, entertainment, information, etc. This kind of a classification is best supplied from analyzing the specific services and media customers' use. The second step is a detailed analysis of previous data usage patterns to generate usage profiles according to time of usage, such as day/night, week-end/weekday, or working hour/leisure users, according to usage frequency and recency, such as frequent/rare users. Customers' needs and behavior, along with handset capabilities and geographical presence, will also help determine the untapped potential for incremental usage.
For new customers, the critical indicators of potential will be their mobility and their geography, since there is no previous data usage knowledge about these cust-omers. In addition, usage patterns of relatively traditional telco services such as voice and SMS will help the operator predict the need for data services. Also, by estimating customers' location information and change in location it's possible to identify their likelihood to use mobile data/Internet services, build target groups using that information, and then to introduce or promote data usage to those groups. It's important to emphasize in the campaign that the customers do not need the data itself, but the service being used by the data, such as social networks, Web browsing, or email. Therefore, the service offer should be based on the potential needs of each identified target group.
Many customers spend long hours commuting to work, for example. If they take public transportation, it is a perfect time for them to read newspapers and check their phone messages. Telecom operators have a large set of data and it is not difficult for them to find these customers in their database. Customers whose phones connect to multiple cell towers on weekdays between 7 a.m. and 9 a.m. represent these types of customers. Imagine a marketing campaign like this: "Do you want to read newspapers on your mobile phone? Simply pay $5 per month." This is more relevant and gives more context than a generic marketing message like, "Purchase 100 MB data package for just $5." Marketing based on customers' behaviors and needs enables operators to target the right customers at the right time with a relevant value proposition in an understandable way.
The above examples are only an introduction to this new dynamic approach to analytics. The opportunities to use transactional behavioral analysis and capture data potential are endless, and doing so can add millions of dollars of added value for operators. The balance between dynamic analytics and business perspective, supported with an effective execution mechanism, will be the new frontier of operators' future growth.
