We are dedicated to helping our members achieve their health and fitness goals. We understand that sometimes life gets in the way, or they might hit a motivational wall, especially this time of year. That’s why we are embracing an innovative, proactive, new approach to member support and retention.
We are now utilising a self built sophisticated machine learning model that is able to predict which of our members might need our help the most. Machine learning is a subset of artificial intelligence where systems learn from data without being programmed. Our machine learning model has used data from a total of 254,963 active and lapsed members (from 1 July 2021 onwards) to identify which of our current members are displaying similar patterns of behaviour to those that have cancelled their membership previously and therefore might be in need of our help. This allows our teams to reach out and offer support.
Our machine learning model considers 17 different factors of behaviour to be significant in predicting those members that need a little extra support or guidance. Across a dataset of 90,000 active members, the model predicts that 32% are at risk of simply cancelling their membership because they’re potentially not getting the support they need. Crucially, this model has achieved an 84% precision rate and so we have high certainty in its predictions.
We’ve been able to do all of this as a direct result of utilising the Google Cloud platform, using solutions such as Google BigQuery as a powerful, fully managed, serverless data warehouse designed for analysing large-scale datasets.
We’ve built a BigQuery machine learning boosted tree model, a type of machine-learning model that’s good at spotting subtle patterns in member behaviour. Using BigQuery’s built-in machine learning functions we can turn raw data into clear risk scores that our teams can act upon. These functions are built into Google BigQuery so they give us the flexibility to experiment and improve our model quickly, without needing complex additional tools or infrastructure.
If the model identifies a member as needing extra support, it’s able to prioritise the need accurately.
Our calls are designed as a friendly "check-in" to offer support and encouragement, nothing more.
Depending on the member’s specific needs, we can offer solutions like:
| Solution | Use Cases |
|---|---|
| Programme Review | For members with a lack of time, lack of motivation, or not currently seeing any progress towards their goals. |
| Refresher Session/Re-Induction | For those lacking confidence or unsure how certain equipment works. |
| Boditrax Scan | For members lacking motivation, not seeing progress and with specific body composition goals they want to achieve. |
| Queue Jump Class Booking | If you are having trouble booking a class or are lacking motivation to exercise alone. |
| Bring a Friend 3 Day Trial | For members lacking confidence in the gym on their own, not enjoying the gym, or feeling intimidated by the environment, bringing a friend or family member can really help. |
This new approach is entirely focused on providing a better service and demonstrating support for members who may be struggling. By proactively helping them stay engaged, we know they will be more likely to achieve their fitness goals.
This is currently an ongoing trial at 5 very different sites across our 3 regions. The efficacy of our new proactive support approach will be closely monitored over the next few months, during which time we will be assessing which of our retention tools are most requested during calls and support our members the most. The plan, if the early success we’re seeing continues, is for a wider rollout in early 2026 - across all of our remaining sites.
We're excited to offer this higher level of support and look forward to seeing more members enjoying their experience with us!
