As modern analytics offers a granular understanding of the customer journey, marketing and retention teams now have a wealth of valuable customer data around usage, transactions, and subscriptions. Once collected, this data can be used to increase retention by allowing teams to discover churn levers, predict customers at risk of churning, and decide the best strategy for lapsed users. With this knowledge, teams can take tailored actions for at-risk users, win churned users back, and understand why users are leaving.
Although rules have traditionally been used in retention, to truly take advantage of this dataset, companies are adopting data science models as a central component of their retention strategy. These models allow teams to take a tailored, per-customer approach based on a vast amount of different data points which they already have.
The right data science retention strategy depends on your business model.
In this scenario, a business is trying to predict which customers are likely to churn on their subscription. With many subscription businesses, transactional data may be the same from month to month (i.e. you have a single subscription fee), so the most commonly used variables are based around usage. The best usage data is that which varies over time, such as page views, product interactions, emails opened. These data points can be fed into models which find patterns in historical data from users that have previously churned, and use these patterns to 'score' current subscribers on the probability that they are still 'alive'.
Although scoring whether a customer is alive or not can be powerful, to next obvious question is what action to take for an at-risk customer, as customers may be deemed at-risk for many different reasons. This means it's important to understand, for each subscriber, which input usage variables resulted in the at risk score. With this tailored output, companies can take an action to increase that input and thus the health of the customer.
E-commerce or retail businesses typically rely on repeat purchasers, and try to predict the probability that a certain customer has churned (i.e. won't purchase again). This can be accurately predicted using historical transactional data on purchasers, and data science models can score customers individually based on their personal historical data and that of the entire customer cohort. Traditionally, very simple rules been used to used to decide whether a customer is still 'alive', with businesses using the same rule for all the customers. However, to increase retention, rules must usually be set on a per-customer basis. For instance, consider a customer who bought from you every day for three weeks straight, and we haven’t heard from them in months. What are the chances they are still “alive”? Pretty small. On the other hand, a customer who historically buys from you once a quarter, and bought last quarter, is likely still alive.
Companies often use the output of this model to target winback offers to at-risk customers, without cannibalising margin by sending offers to customers who are active, but infrequent, purchasers.
How can NStack help increase retention?
NStack provides two retention models which have been customised for each of the business models and retention strategies above. These models can quickly implemented by in a self-service manner by commercial teams or data analysts, without requiring a specialised internal team or consultancy. The NStack platform automatically validates the outputs of these models to ensure you can be confident in the results.
NStack Subscription Retention Model
This model lets you predict which subscribers have a high propensity to churn, by inputting usage data and subscription information. Output includes customer risk score, and weighting of input variables (i.e. which inputs contributed to health) so a custom action can be taken.
NStack Transactional Retention / E-Commerce Retention Model
This model uses transactional history to predict the probability that a given customer is alive, based on their historical purchasing patterns and the patterns of other customers.
How do I use these models?
NStack provides a web interface where you can connect models to your data by building drag-and-drop workflows. This process typically takes 10-20 minutes. These workflows allow you to take data from various sources, such as CSV, or your data warehouse, run data through the model. The model can output data as a CSV, or can automatically write to warehouses, CRM, or email marketing platforms. These workflows can be scheduled to run automatically, which allows data science to become a central part of your workflow.
Most frequently, customers use the output of retention models to take some form of action. In e-commerce, this might be sending different levels of winback offers to at-risk or churned customers. In many subscription businesses, this may be taking a custom marketing action dependent on what the churn levers for a particular customer are. In B2B SaaS businesses, this may be setting a task on Salesforce and contacting the customer directly.