Customer Retention


Customer Retention

Customer retention is an end-to-end solution, helping reduce customer churn and determine the cost of retention, prepare a retention dashboard or set up a reactive or predictive retention using the PtC, PtS or CLV models.

The customer churn rate is a problem primarily in the highly competitive environment in saturated markets (finance, retail, energy). Many companies nevertheless underestimate the seriousness of this problem or incorrectly focus on the retention of unprofitable customers. There is, however, a verified and analytics-based road to success.


Thanks to many completed projects in the area of retention, we have gathered knowledge of all factors that influence the retention of the right customer and retention success, and we have integrated this knowledge in the solution focused on customer retention. Our solutions are always adapted to specific client needs and may include management of retention teams, performance indicators, retention offers, predictive retention models and reports. We provide retention on an end-to-end basis and we have repeatedly helped many clients optimise performance and increase the focus on growth.

Thanks to proactive retention using the Propensity to Churn model we decreased the churn rate of one of the largest Czech banks by 41%.

Other areas of use of retention

Cost of retention

We often encounter situations that companies are unable to assess how many customers they lose every year and how much money it costs them. This assessment is not trivial (although it may seem so), but it is a necessary first step to a well-managed retention process.

Retention dashboards

The customer churn rate must be not only measured but also regularly reported, trends must be measured in order to assess the efficiency of the retention process. This is what retention dashboards are for. We can design them, prepare data from them, create, deploy and train.

Competitor model

One of the most significant churn rate factors is the competition. It is necessary to know the competition and to identify the Company’s strengths and weaknesses in comparison with the competition. The competitor model calculates for how much a customer would get a particular product or service from the competition and how much they would actually pay for it (total cost of ownership) in comparison with what they really paid at “their” company. This calculation is a basis for smart customer communication, where arguments along the lines of “all that glitters is not gold” supported by numbers gain amazing persuasive powers.

Reactive retention

The first step on the long road towards efficient retention is usually the process of reactive retention, i.e. how to work with customers that are already about to leave or are threatening to leave. We look at the customer journey to see if the process covers all the leaving customers, on retention offers, on profitability by segment and offer and on communication channels. The typical output is usually the introduction of unified governance, enhancement of the range of retention offers, data consolidation, change in directing customers to offers and elimination of unprofitable offers.

Predictive retention

Predictive retention (or proactive retention) focuses on customers who have not left and are not threatening to leave, but many of them still are at various stages of the decision to leave. The goal is to identify the ones most at risk and select a strategy that will eliminate the risk of leaving as much as possible for the lowest possible cost. The key instruments are predictive models of propensity to churn (who will leave?), CLV (how much money would I lose?) and propensity to save (how difficult is it to convince them to stay?). It is necessary to realise that the risk of leaving does not concern only long-term customers, but the danger exists from the very beginning of the customer relationship. The predictive retention process therefore pervades the entire CRM process and should be well integrated therein.

Propensity to Churn (PtC)

PtC is a predictive model that is able to predict the risk of a customer leaving. It is built either as a binary classifier or as a survival model. It is a technical component of retention solution and it is not functional on its own. In order for the PtC model to work well, it is necessary to implement it in the client’s environment, connect the output to retention offers, deal with governance, set up reporting, train users and ensure regular maintenance of the model. We often come across situations where the client tried to develop a PtC model in the past without much success. Our recipe consist in correctly defining the modelling target and linking the output of the model to the execution of the retention campaign.

Propensity to Save

Propensity to save (PtS) is a predictive model that predicts the success of a retention event at the level of individual customers. What we generally perceive as success is when the client is saved, i.e. does not leave and stays active and profitable after a set period after the retention event has elapsed. PtS-type models have so far been something completely new for all our clients home and abroad, so we do not think it presumptuous to believe that Deloitte Advanced Analytics was the first in history to name, create and deploy this model. A PtS model is typically simple (only a small development sample is needed for its development) and must be updated quickly (the retention process changes dynamically throughout the project).

Customer Lifetime Value

The retention programme segments customers based on their value, propensity to churn and propensity to save. These three key indicators create the business case determining which retention strategy and which offer we will use for the particular customer. Customer value is based on the Customer Lifetime Value (CLV) model. Customer value is defined as the present value of all revenue (or profit) from the client over the period of the next several years (e.g. 5 or 10 years). CLV is the result of fine segmentation combined with Markov chains. Just like PtC, CLV is a technical component of the retention solution that needs to be wrapped in the process and execution, but unlike PtC it has a wide range of uses outside the retention area. Learn more.


Retention governance

The basic condition for any retention process to work is that is has to be owned. And that it has to be owned only by one person, not by everyone and just a little bit. Retention governance is nothing more or less than an organisational change that prepares the company for efficient operation of the retention process.

Call centre analytics

The call centre is an important and indispensable communication channel in the retention process. As part of the retention programme we had to make many changes in our clients’ call centres. For example, to create a retention team, set up motivation, capacity model, information flows, prioritisation and direction of customers to the individual retention events and operations, coordination with the branch network. Since this is an extensive area and we have already gathered significant experience, we now offer call centre analytics as a separate solution. Learn more.


Retention offers

Advantageous savings, cash back for payments at retailers, preferential treatment of payments for energy sent from current accounts, lowest instalment guarantee – these are just some of the examples of innovative retention offers that are attractive for customers. From the long list of offers, the best ones have been distilled using customer surveys and practice shows that they have really shaken the old-fashioned world of free accounts.

Churn triggers

How should we even communicate with a customer who wants to leave? What do we know about them and what of it can we use during communication? Churn triggers are customer events that increase the likelihood that a customer will leave. These events are data-defined, regularly recalculated and saved in a database and paired with the customer profile, which serves as basis for the call script. These triggers offer the operator ammunition for leading an efficient conversation with a customer and increase the probability of saving the customer.


Filip Trojan

Filip Trojan

Senior Manager

Filip is a Senior Manager in the Advanced Analytics deparment. He has over 15 years of experience in analytics, machine learning, mathematical optimisation and data science. He has an extensive variet... More

Veronika Počerová

Veronika Počerová


Veronika is a manager in the Advanced Analytics department. She specialises mainly in analytical end-to-end solutions for clients from the finance, energy and retail industries. She focuses on predict... More