Our Machine Learning solutions identify potential customer churn as early in the customer journey as possible and give you the tools and insights to proactively manage the customer experience.
Managing churn is a major growth driver for any business. Optimising the touch-points of your brand across the customer life-cycle to ensure they experience the best of what you have to offer is critical to underpinning the resilience of your business. The more you grow, the smaller the available pool of prospective new customers become, and the more challenging new acquisition becomes.
Investing in customer retention will benefit you in both the short and long term. Not only does retention grow your overall value, add to this the opportunity for broadening sales and upselling new products and services, and it makes the business case for a comprehensive retention plan even stronger.
Churn rates vary widely across industry, but in sectors such as Telco, Financial Services, Travel and e-commerce it is perfectly usual for it to be running at between 10%-20%. Consider the business with a relatively average churn rate of 15% compared to a peer with say only 5%. After five years based on churn alone the high performer will be more than 1.5 times bigger than its peer.
Acquisition has long been the keystone to a high-growth plan. But consider the business that experiences the double whammy of driving growth through customer success and expansion via reduced churn and that goes on to use this knowledge to continually optimise its acquisition approach to refine and perfect the targeting of new customers.
Causes of Churn
Learning from churn is the key to running a successful business and our solutions enable our clients to continuously analyse and understand the different types of churn they are experiencing and what are the right actions to take.
Some churn can be considered natural and hard to avoid because those customers should never have been customers in the first place. Perhaps they have been targeted by a compelling offer, or a complex deal and persuaded to sign up without really understanding what they were buying into. This is damaging on two levels. It is wasted investment in acquisition costs and potential commissions to affiliate partners and it does very little to build brand equity or reputation.
Other types of churn are avoidable and typically fall into the following broad areas:
On-boarding challenges - Set-up delays, process hiccups and ineffective product education lead clients to give up and look elsewhere.
Fee/Product/Service shock - Client realisation of the deal they have signed up to post the initial welcome period causes them to cancel.
Weaker ‘Promise’ - Mid to longer term clients have fees, products, services and charges applied that do not suit their use or don’t meet a more promising proposition from a competitor, so they start to look elsewhere.
In order to understand both natural and avoidable churn we start by understanding what customer success looks like. We identify customers by their success with your product be that through their tenure, engagement, breadth of use/adoption of the product, growth into other services and what they have experienced from the brand along the way be it through customer service experience, marketing, product experience or fees/prices applied.
Build understanding of your attrition.
Identify key triggers and available actions.
Set baseline attrition measures.
Define your client plan.
Build machine learning models to proactively spot triggers and predict attrition early enough to act.
Build data triggers and profile variables to drive actions.
Establish ongoing test and learn to optimise proactive and reactive attrition strategies.
Ongoing model optimisations and learning.
With this we can identify the points in time when churn is happening and the likely drivers behind it. It will also tell us what triggers we could have seen in the data and through test and learn what are the best actions we can take to mitigate or reverse the churn.
This data driven understanding of the reason for churn, the signals that indicate how likely it is to happen and the potential remedies provide us the necessary ingredients to build the Machine Learning models to continuously track individual customer experiences and reactions and predict with ever increasing accuracy how likely they are to churn.
Combining this with the understanding of what makes a ‘good’ customer and where to find them we can begin to devise the ideal customer experience and journey to suit each and every individual customer.
This means from the moment we acquire a new customer we can adjust the key experiences to align with their current progress on the journey and ensure we do not inadvertently trigger the desire to churn. For example; if on-boarding has not gone as smoothly as planned and the customer has taken longer to start using the product we can take smart decisions to rectify this with extra servicing, additional training or if necessary extending the welcome offer to reflect their experience.
This enables our clients to move away from a reactive state to one that is proactively enabling the development of a ‘customer relationship’ built on an intimate appreciation of each customer. This allows them time to build trust, repair mistakes and build strong ties. This is the foundation of building value in your business. To find out about more insights on business go to our featured insights pages.
Featured Case Study
Using Random Forest machine learning we developed a model that continuously looked at individual cards behaviour and alerted if action required.
A tailored attrition model for each individual customer spots attrition earlier and value at risk modelling prioritises the attrition activity. Automation both increases speed of action to remedy attrition and boosts productivity of agents.
Credit Card Attrition Modelling Outcomes
Reduction in lost customers through call centre outreach.
Attrition in the customer base had been steadily growing as a result of increasing price competition and a number of customer experience issues.
Current methods for identifying and evaluating attrition proved to be very blunt and did not factor in normal and seasonal behaviour.
Customer agents were making many calls to customers that were not at risk of leaving and too late in the customer jounrey for those that were.