Organizations engage with their customers via multiple channels. In saturated and highly competitive markets like telecommunications, carriers are investing heavily to minimize churn by delivering, personalized customer interactions through these channels. Customer satisfaction is contingent on the telecom providers’ ability to present personalized offers through these customer contacts.

The leading recipe for delivering the ideal “best next offer” involves a collection of technologies and processes leveraging data, analytics and optimization. Customer-level behavioral data across channels need to be harnessed, processed and made available to generate insights to predict future customer behaviors. Predictive and, ultimately, prescriptive models enable the marketer to project the efficacy and success of each offer. This insight is then fed into a decisioning process, to organize the offers in a prioritized sequence.

This decisioning platform typically involves rules-based and/or self-learning algorithms that compute prioritized sequences of offers for each customer. These sequences are consumed by digital marketing systems that present offers through online and offline channels. For a customer logging into his or her account, the execution system can surface a customized offer by extracting it in real-time from the prioritized sequence already in place for each customer. This kind of prescriptive model takes into account customer propensities and contextual information across online and offline channels.

Organizations have realized significant lift in customer responses and loyalty by basing their decisioning on rules or predictive analytics to generate prioritized sequences of tailored offers. But marketers can do even better, realizing significant gains in ARPU and customer lifetime value, and decrease in churn (through upwards of 20% reduction in costs, 5-10X improvement in responses and more),  for both the customer and the organization by incorporating true optimization in the Best Next Offer decisioning system.

Most organizations have a plethora of customer and transactional data, and have begun to apply the data they have to a rear-view mirror analysis of trends and investment in predictive analytics to guide future offers or campaigns.  Some are now ready to graduate to adoption of true optimization.

Optimization is a game-changer when there are choices to be made, and the financial, operational and other business level constraints on those choices are known and need to be satisfied. Making these choices requires the organization to know the relative benefits of each choice for each of its recipient. The customer data required for such optimization includes what choices have been made in the past, and what sequence of offers were made based on those choices. Records of those decisions generally can be found in the company’s transactional systems.

Some of the data required for decisioning is available for analysis in advance. If, for the required use-case, that data is sufficient, developers have the luxury of creating an optimization model based on data that already exists.  In interactive channels, however, some of the data is generated at the moment of customer interaction. The model then needs to be able to capture the new data and incorporate it into the decisioning process, at the moment when the best next action is prescribed resulting in optimal real-time personalization.

An optimization strategy is computed based on historical data consisting of decisions made on all customers, and the objective desired and the constraints to be satisfied. The strategy is then surfaced in the execution system,   waiting in real time to be called for each new customer interaction. Data variables are passed to the executable to compute the unique Best Next Action for that personalized interaction.

A combination of online and offline data is used feed the optimization strategy. The data includes individual customer and his or her behavior, what the customer was offered in the past, and what the customer accepted or rejected, along with analytic model scores.

In real time, the customer’s contextual behavior is captured – e.g., what content is the customer consuming on the web site? How much time is he or she spending on specific pages? What kinds of content is the customer asking for? What data has he or she provided? This real-time data, along with offline data, is fed into the optimization logic to arrive at optimal next-best recommendations.

Typical best next offer personalization systems currently base their decisions not on an individual level but on a persona – a psychographic profile of a composite customer. Each individual, rather than being considered uniquely by the model, is assigned a representative persona. The organization may use sophisticated analytics to determine which persona best applies to the individual. It is less resource-intensive and easier to base decisioning on personas than individuals, assuming customers are assigned appropriate personas.

Still, while the market is gradually becoming more comfortable with persona-level optimization, it is widely perceived that true personalization of the customer experience requires individualized data. The market has now evolved to enable telecom providers to achieve true personalization and gain substantial improvement in ARPU, reduced churn and costs by optimizing at the customer level.

Is your organization ready to realize significant gains through the application of Best Next Action optimization? One expedient may be to start in one line of business or product area, and then extend adoption across the business as you gain successful experience and measure the uplift. Understand further to how this is possible through FICO’s customer brief on Best Next Action for Telecommunications.

For more insights on this type of optimization, please register for our October 14th webinar “Optimized Best Next Action Solutions to Grow Customer Lifetime Value.”

Leave a Reply