Mitigating value at risk is not only about measuring the propensity of unsatisfied customers towards churn but also to identify the respective constituting factors. The ultimate value at risk is then the lifetime value of those unsatisfied customers. This activity constitutes of the following three use-cases fueled by several models.

Customer Lifetime Value (CLV)

CLV is one of the critical KPIs that helps organizations to measure the value of their customers. We employ several models based on recency, frequency, and monetary value of the past purchases and forecast the estimated lifetime value of the customer.

Propensity to Churn

From the past behavior of churners, find the customers matching the churn patterns based on their recent individual behaviors and experiences. Use A/B testing to validate the hypothesis and perform out-of-sample validations to ensure consistent results.

Deduce Actionable Insights

The high probability churners are provisioned by personalized retention offers. Additionally, it is required to gain actionable insights on major factors governing churn, e.g., mediocre network quality, bad customer experience, bill shock, new handy, etc.

Every year, network operators invest several million € in the network to ensure best in class customer experience. Network rollout is a high budget activity, which if mishandled can lead to great loses due to investment in wrong locations. We therefore, provide extensive analytics to optimize this process. This activity consists of the following two use cases fueled by several models.

Quality of Experience (QoE)

We provide several models to help measure the QoE of network users. QoE measures the content delivery which refers to the acceptability of an application or service as perceived subjectively by the end user. With the help of a customer experience management (CEM) system, QoE models convert customer sentiments into scores from 1 to 5, thereby converting the psychological aspect of customers into a measurable quantity.

Value Based Network Roll-out

QoE is now the target variable that need to be optimized by the network roll-out plan. Several graph theory based models are built in order to provide two distinct optimization plans, chosen as per the able CAPEX. I.e., an upgrade plan to ensure x% of end-to-end customer satisfaction, or a plan to guarantee x% of service satisfaction.

Network operators have to stay vigilant in order to keep their customer base up-to-date as per their current front-book tariffs. In order to ensure that, operators are needed to be continuously engaged in migrating their customer base to the next best up-sell offer. This activity constitutes of the following two use-cases to help operators in continuously increasing their base value.

Next Best Offer

Our consumer behaviour models already provides a good lead on customer preferences and when further joined with demographics data, their predictions reaches a great deal of accuracy in terms of predicting the most likely target for pitching any up-sell offers.

Best Channel

Based on customer engagement on various channels and his past behavior w.r.t purchase orders, we offers models to identify the most suitable channel w.r.t customer preferences. This also includes models to sketch customer journeys, measure the contribution of individual channels, and the interplay of those channels to maximize convergence.

Unlike up-selling which is about tariff upgrades, x-sell strictly aims to increase product count per household. This activity is broken down into the following three use-cases fueled by several machine learning models.

Identify the right customer

Learn from the interactions and transaction history of converged customers, perform a lookalike analysis on current customer base to find customers following the similar track of convergence.

Identify the right product

Using association rule mining techniques, learning from the history of combination of various tariffs and recommend the tariff that best fits with the current tariff plan of the customer to whom the above model selected for x-sell.

Identify the best channel

Based on customer engagement on various channels and his past behavior w.r.t purchase orders, we offers models to identify the most suitable channel w.r.t customer preferences. This also includes models to sketch customer journeys, measure the contribution of individual channels, and the interplay of those channels to maximize convergence.

The ever evolving market competition results in the accumulation of out-dated tariffs in customer base which further hinders several learning processes, e.g., association rule mining to recommend most likely product a customer will buy. It is therefore necessary to define rules to map out-dated tariffs to the latest front-book. Additionally, find opportunities w.r.t new products to gain any competitive advantages. This activity is broken down into the following two use-cases.

Tariff Mapping

There is always a portion of customer base with out-of-date tariffs. In order to find the right product for the customer, x-sell models are trained on past data which also includes old tariffs (tariffs no longer actively sold) and therefore gets recommended for the new customers. In order to mitigate such out-dated recommendations, mapping models are required to sketch a one-to-one relation between old and latest front-book tariffs.

Tariff Optimization

The stringent competition in telecommunications market results in every decreasing price per unit GB. It therefore becomes a necessity to regularly inspect current front-book tariffs in order to find any potential gaps which might lead to tangible competitive advantage.