Do you have a high, or growing, number of churning customers? Does your company struggle to pinpoint the exact drivers of customer churn? Is your ability to anticipate churn insufficient, or does it need improvement? Do you see limitations in your ability to accurately measure customer value over their time with your company? Is putting together a single view of your customer a challenge?
Some industries see average customer churn rates as high as 30% within a year. Markets are saturated with competing offers and services, unhappy customers are quick to leave or downsize their service usage, and a class of “professional churners” is beginning to emerge. Given these challenges and the ever-increasing competitive pressures, managing your customer base to reduce churn should be seen as a top priority for any business.
Winning and attracting new customers is a cost-intensive affair while maintaining happy customers often comes at a much lower cost. That’s why it is crucial to focus on identifying at-risk customers and designing an optimal retention strategy to ensure long-term stability for your company. Unlocking this requires precision understanding of your customer base, relevant and tailored offers fit to their context and needs, and appropriate engagement via the right channels. Customer retention analytics is the practice of using data to precisely understand your customer base and determine the appropriate action to save them. It is a key enabler for maintaining a stable customer base at solid margins, staying competitive in the market, and alleviating internal cost pressures. It’s much easier to retain your existing customers than to attract new ones. Customer retention analytics can help you retain high-value customers and identify their reasons for sticking around for the long term. Access to customer retention and customer satisfaction data can help your business make key developments to drive business growth and service innovation. This approach can then be coupled with internal innovation and product/strategy development to further enable continued long-term profitability.
Since the cost of acquiring a new customer is higher than retaining existing ones, it becomes imperative to determine when and why a customer might leave. Without a focused and data-driven retention strategy, the knowledge of why customers churn is delegated to expert judgment alone. This puts an immense challenge on institutionalising that knowledge, de-risking it from the attrition of key employees (experts), and hard to verify and validate.
By pulling together your available customer data and harnessing the power of multiple complementary Machine Learning models, we can help you examine the factors driving customer retention, gain precise insights and create actionable strategies to reduce future churn.
Our approach to retention analytics aims to retain existing customers at optimal cost and to prevent them from exiting to competition. It is designed to extract maximum insight from the data available by minimising the false positive error (predicting a customer as a churner when they didn’t) and the false negative error (predicting customers not churning when they did churn).
By pulling together a diverse variety of your datasets, we can help you: save profitable customers, reduce your total retention costs, precisely define your up & cross-sell targets, improve customer experience, and help you institutionalise the knowledge gained from these insights We have designed our solution with transparency and explainability built in to provide continuous alignment with your business goals by:
- Having the number of customer predictions delivered each day tracked to ensure that measurable outputs are provided;
- Providing probability estimate for each prediction to ensure only high-quality ones are delivered;
- Showing the Customer value with each prediction for transparency and ease of prioritisation;
- Sharing periodic reporting on the outputs we have delivered allows for two-way dialogue between your business teams and our expert data scientists.
We have broken down our approach into 4-steps, which, through our experience, appear to provide the best balance between a standardised approach and a custom, fit-for-purpose consultation:
1. Business Objective scoping
We capture your business objectives and challenges in detail and set up goals for the project. We inventory all your current retention programs, their effectiveness, and the underpinning analytics. Then, exploratory workshops with key stakeholders are organised to understand your business activities and past, current, and future data availability.
2. Modelling and Analysis
Three meaningful types of data are needed to form the foundation for a comprehensive customer retention analytics approach. We collect Customer Behaviour, Outcomes, and Intent data from your source systems and relevant databases. This data is used for descriptive analysis and as a proxy for future customer action. The technical approach begins with understanding the existing customer base by segmenting them into homogenous groups. The choice of features for segmentation is critical. We ensure that the chosen features represent distinct concepts and/or patterns. For example, we can consider these as features representing a transaction, behaviour, intent, demographic, and/or firm characteristic. Separately, we calculate the value each customer is likely to have over their long-term stay with your business. This is typically referred to as the Customer Lifetime Value (CLTV), which is calculated as (generalised formula):
CLTV= (E[Transactions]* E[Order Value]) – (Cost of Acquisition+ E[Cost to Serve])
E[Transactions] – the expected future number of transactions
E[Order Value] – the expected value of future transactions
Cost of Acquisition – the actual cost to acquire the customer
E[Cost to Serve] – the expected future cost to provide service
The three “estimated” components of CLTV, the number of future transactions, the order value per transaction, and the cost to provide service are predicted using probabilistic models. They typically depend on recency (how recently a customer has made a transaction), frequency (the number of times transactions have been made by the customer) and monetary values (average transaction value by the customer). The probabilistic models are sometimes defined for each segment or relevant sub-group to improve their precision. After testing and validating the models over an 8-week period with live (new) customer data, they go live by being integrated with your downstream operational systems and business processes. Then, we can start providing daily information delivery on each customer relevant for retention, with a score to advise whether rescue them or allow them to churn. Depending on your business objectives, customers can further be ranked based on their propensity to churn to prioritise rescue efforts. For example, customers with a high probability to churn and a high lifetime value are highlighted over others for faster rescue efforts. The specific nature of the rescue efforts is informed by recommendation models for cross-sell or up-sell.
3. Operational Integration
We integrate the retention analytics engine into your downstream systems and operational tools you use within your customer support teams (e.g., queue tools and agent dashboard). We collaborate with relevant internal teams, like IT, CyberSec, BPO, and others, to establish secure data pipelines and integration. Through experience, we’ve learned it’s critical to stay closely connected with the relevant infrastructure and operations teams as the business IT environment is constantly evolving. That is why we also set up an ongoing check-in to ensure continuous integration with our solution and reduce the risk of potential future disruption.
4. Daily Data Delivery and Operation
The retention analytics solutions are now integrated into the operational tools of the client, ready for daily use. Every day we deliver a new list of individual customers at risk (e.g., in list form) to your customer service teams who can reach out to the customer and undertake the rescue operation. In the background, we also monitor the data science model to ensure that it performs within agreed norms (e.g., SLA).
Reducing churn through predictive NPS for un-observed customers
increase in retention
incremental value in 1st year
”Working with Amplify Analytix in this effort helped increase retention successes by 34%. The daily feed into our outbound call management system made it really easy for the team to use. No new dashboards or platforms to be trained on![Customer Care Team Lead]
Reducing churn is now more important than ever, especially considering the increasingly competitive market dynamics in the post-pandemic inflationary economy. Yet many companies have not taken the steps required to build a solid data foundation to underpin all customer-facing activities. А focused and data-driven retention analytics approach can help you identify your most volatile customer journey points and effectively understand the root causes that motivate customers to churn. With such insight on hand, you can quickly and precisely target these customers with targeted marketing efforts and churn prevention strategies, engaging the right customer at the right time with the right message.