After 18 years working with some of the biggest names in Fintech, including Thomson Financial and SS&C, Ian Mullane left corporate life as the COO of Sungard’s Asia Pacific business. Since then, he’s founded and grown a number of companies including Locowise.com, Vanda.fit and most recently, Keepme, an AI-powered membership retention tool. We sat down with Ian to learn more about:

• The current state of retention in the fitness industry

• How AI and machine learning can help the health and fitness industry improve the annual member attrition rate, and

• How operators can effectively incorporate AI into their business

What is the current state of retention in the fitness industry?

Very little has changed over the last decade. We have access to ample research about the factors that impact retention, but as an industry, we have not managed to significantly improve retention rates. Despite all the research available, we’re simply not seeing the decline in attrition you would expect.

Generally, the accepted industry average is that 40% of members will leave a club on an annual basis.  Increased competition, a gradual slowing down of the gym penetration rate and the current economic environment will all impact retention so we need to get better, quickly.

How are operators responding?

Every operator will tell you that retention is important. However, despite all the suggested solutions, very few have made progress. Many operators make member retention the central pillar in everything they do, but new membership sales often cover up the impact of existing members leaving. Therefore, retention isn’t a problem they prioritise because overall, the numbers look good. But imagine you maintained your new sales pipeline and retained just 10% more of your existing members? Think about how much more profitable your business could be.

If you hang on to more of your members, they are going to be far more valuable to you over time. They will refer and recommend you, which in turn, attracts more people. This is necessary for growth. Relying on new sales alone is not sustainable and penetration will not rise forever.

How can AI help solve the retention problem?

Firstly, it’s important to note that AI alone cannot solve the retention problem. What it can do is provide operators with the insight and confidence to take action. It will then show them how that action is yielding results.

Right now, we don’t have insight into which members are likely to leave us in a timeframe in which we can do something to change it. AI can provide this insight and give an operator a greater understanding of their overall attrition risk. Is it group exercise? A particular membership type? What gender poses the greatest risk? AI can provide this insight in enough time to allow an operator to do something about it.

Can AI replace the role of human intervention?

AI should not be viewed as a human replacement. It should be viewed as something to support existing staff. For example, imagine you’re a PT or a member of the floor staff. You can see there are 60 people in the gym. You know that personal interaction can play a big role in retention but speaking to all 60 people is time-consuming, and for some members, unnecessary. In an AI environment, you can tell immediately which people are at risk of leaving and would benefit most from an interaction.

Does the deployment of AI require a staff of data scientists and analysts?

Ten years ago, yes. There weren’t the tools available to put AI predictions into a business context. That has changed. Today, all you need is someone who knows what they want to predict, and what they will do once they get that prediction.

What is the one thing that a fitness operator should consider prior to deploying AI?

Look at the data that exists within your business and understand that the value of that data will increase exponentially over the next few years. More and more tools will be able to deliver profit-driven predictions based on data. Make sure you have a proper CRM system because prediction accuracy reduces if your original data is inaccurate or incomplete.

Greater emphasis needs to be given to gathering and managing this data. Operators currently collect data for a specific purpose – ie: age, name, address, bank details - but they also need to understand what other data an AI model would find useful. The effort and investment made now could generate millions of pounds in the long term.