Making the Most of Your Data: Using the Keepme Score™ to Retain Your Members

Making the Most of Your Data: Using the Keepme Score™ to Retain Your Members

When thinking about any membership based business, they share a common pain point in member retention. Most fitness businesses predominantly run on memberships and are therefore acutely aware of this pain point. Retention matters, there is no question that retaining your members has a positive effect on the business’ bottom line, when we consider that client acquisition can cost between 5 – 7 times more (Forbes) than retaining existing clients (depending on the sector). So, with this said, why is it so difficult for fitness businesses globally to retain their members? 

There is some difficulty inherent in the sheer number of different factors that can feed into why someone might leave as a member. Some can be controlled by the fitness business i.e. price point, facilities, classes, additional benefits, and communications. Then there are other external factors that cannot be controlled, i.e. change in a member’s circumstance, or moving away from your location to name a few. However, when it comes to those factors that can be influenced by the fitness business, it is imperative that the right strategy is implemented to ensure the best results.

In this blog we illustrate how using the right combination of data (data that already exists in your member management systems), The unique Keepme Score™ can support a membership retention strategy like no other.

Membership Retention Strategies are Key 

The most important thing is that you have a membership retention strategy in place. Most fitness businesses will have them in some format, some will be formalised and others less formal. Aside from a good product (facilities, classes, equipment, and support) at a good price point, you want to build a sense of community. 

There is plenty of data stating that having regular contact with your members will support retention. An interesting statistic is that “Two interactions a month between staff at a health club and their members can reduce membership cancellations by up to 33%” (RunRepeat.com). This illustrates that you need to have a robust engagement strategy as part of your membership retention strategy.  

Most fitness businesses know what they want to say to their members, but it's hard to know who to talk to and when. For example, you can send a milestone email, but what about those in-between moments, like when a member finishes their 20th class but might not come back after the 30th? What if there was a tool to help identify members who might leave? 

That's where Keepme and the Keepme Score™ come in. It's a unique metric that uses data and machine learning to predict if a member might leave. Let's see how it works.

The Keepme Score™ is a unique metric for each client. By leveraging member data and machine learning, specifically a Random Forest Classifier model, Keepme can predict the probability of a member leaving and assign a score to help decide the appropriate action.

Let's look more closely at how this score is calculated. 

👉 If you haven't already I recommend you dive into our article: Life After January: A Data-Driven Approach to Preventing Dormancy

Understanding the Keepme Score™ Calculation

1. Data Splitting: The member data is divided into 80% for training the model and 20% for testing its accuracy.

2. Training Phase: The model is fed all available data, including outcomes (whether a member will stay or leave).

3. Testing Phase: The model is then presented with the testing data, excluding outcomes, to evaluate the accuracy of its predictions.

4. Impartial Model: The machine learning model starts with no assumptions, biases, or expected outcomes. It evaluates each member's status and considers various data points like gender, membership type, attendance, or newsletter opens.

5. Decision Trees: The impact of each data item is understood through decision trees, which analyse the relevance of each factor to the outcome.

6. Random Combination: The model builds trees randomly, combining them to understand the impact of different permutations — hence the term "Random Forest."

7. Millions of Trees: The model generates millions of decision trees for each member, recording the impact of each combination.

8. Forest Concept: The accuracy of the Keepme Score™ arises from combining the results of all of these millions of combinations.

9. Customized Score: This process ensures that the Keepme Score™ is unique to each member, enabling performance analysis across various groupings like venue, region, or class attendance. 

The factors affecting the Keepme Score™ vary for each customer. You can easily see which members have common factors associated with attrition risk and make assumptions on why that may be before putting preventative actions in place.

What’s more, it’s recalculated daily, incorporating any new data updates from your member management system. Being a machine learning model, it constantly analyzes outcomes, matches them to predictions, and adapts to improve accuracy.

The Keepme Score™ is prominently displayed on the dashboard and individual member profiles, offering real-time insights into member behavior. The higher the score, the healthier the membership base. The dashboard also categorizes members into risk buckets based on their Keepme Score™:

• Low Risk: Members with a high score, indicating satisfaction and loyalty.

• Moderate Risk: Members with a moderate score, suggesting strategic engagement is required to shift them to low risk.

• High Risk: Members likely to leave soon, intervention at this stage is not advised as it is likely to do more harm than good - engaging these members may only to push them to leave sooner.

The sheer volume of data considered demonstrates just how robust the Keepme Score™ is, but also shows the true value of the metric when it comes to implementing a personalized engagement strategy for retaining your members. 

Implementing Automations and Campaigns with Keepme Score™

Utilising the Keepme Score™ is as straightforward as it is valuable, whether in automated member journeys or one-off campaigns. Built-in filters let you set automations based on members’ scores, so you can send appropriate content or offers to the right cohort, and at the right time: 

• The first filter is based purely on score, so you can segment your members based on their score being equal to, less than, or greater than a value of your choosing.

• The second filter lets you segment your members based on % change, so no matter their normal or current score, you can get ahead of any further drop by concentrating on those whose score has fallen by a certain %.

• Finally the third score-based filter lets you segment by status, looking at those Keepme considers to be High, Low, or Moderate Risk.

Using these segments you can send the right message via the most appropriate means, which can of course be automated through the platform. After all, your communication to someone whose score has dropped, but is ultimately still low risk would be very different from your message to a member in the moderate risk cohort, where your intervention is essential to prevent them becoming high risk.

Tracking the performance of these campaigns is simplified, enabling refinement and retargeting based on the data. Keepme takes care of the complex task of segmenting the customer base, making it easier for businesses to strategize effectively. 

Leveraging the power of data and machine learning through the Keepme Score™ provides businesses with actionable insights to enhance member retention. By understanding the factors influencing the score and utilizing it to create targeted campaigns, businesses can proactively address member satisfaction and, ultimately, foster a thriving and loyal community.