Predicting Superannuation Rollouts for Early Action

by Zachary Anstee

In this time of consumer uncertainty, the topic of member rollouts is gaining criticality in the financial services market. Over the years, Tridant has built expertise creating predictive models. Recently, these have been concentrated on the superannuation industry, less as a churn model but rather a ‘rollout’ predictor.

Leveraging advanced data analytics and tools such as artificial intelligence (AI) and machine learning (ML) to better understand – across varied member transactional and data points – member behaviour, predictive analytics enable super funds to be proactive and work, ahead of time, to prevent or reduce attrition.

“Using data, statistical algorithms and machine learning techniques to identify the likelihood of future outcomes based on historical data, the goal is to go beyond knowing what has happened to provide a best assessment of what will happen in the future.”


Rollout statistics are a key metric in retention planning by fund providers. Optimally directed retention planning can help super funds retain lifelong members and save hundreds of thousands of dollars. This makes the business case process for project funding an easy one, as the return on investment (ROI) is quickly clear.

Let’s unpack what the ROI looks like, to feed into and support your business case.

The numbers game

From here on, it becomes a numbers game.

There are several attributes which influence potential savings and I use a couple of assumptions such as number of members and fee percentages, both of which you can adjust to your own company scenario. Let’s start with the first assumption that this is a medium-sized Australian superannuation company with a total number of accounts, or members, of 300,000.

But first, a key learning: There will always be accounts who will rollout no matter what you do. A strong lead indicator to identify these is that they have already made at least one withdrawal (or rollout) or are showing a higher than usual level of activity in their member portal. Those who are starting to rollout are already in the process of moving their money elsewhere, so we might exclude those from our list of accounts. As an average, we find this to be approximately 10% of all accounts, now leaving us with 270,000 accounts to address.

It gets smarter

Next step when refining the list: Factor in the predictive model’s accuracy, and the actionable list size of accounts at risk of rollout. With regards to accuracy, off the bat we find the models show an accuracy of between 18-22%. At first glance, this may seem a little low however we are aiming for small but significant gains to start with, and it’s important to understand that these predictive models get smarter over time. They learn through training thanks to the power of machine learning (ML). Even with 19% accuracy, you will soon realise the impact on the revenue potential.

The actionable list size of accounts at risk of rollout is dependent on how much effort an organisation pours into its retention treatment plan. It would be unrealistic to try to target and retain all 270,000 accounts; budgets and capacity to execute such a campaign would not be feasible. It is up to your organisation to decide a reasonable number, but a common one we see is between 0.25% and 0.5%. Let’s assume 0.35% for the purposes of this article. This leaves us with a list of 945 actionable accounts.

TT8-1

Show me the money

This is where the rubber starts hitting the road. 945 actionable accounts multiplied by a model with a 19% prediction accuracy provides a list of 180 accounts identified with a high propensity to rollout. Our next assumption, and this is a conservative one, is that an average rollout balance is $100,000. The total amount of at risk of rolling out for one month alone is; $17,955,000. The numbers are adding up quickly, that’s nearly $18M per month identified as likely to rollout.

As impressive as that number is, we need to convert it into something more realistic and tangible for a compelling business case. Even with the best marketing and retention plans, you will not be able to save, or retain, them all. Let’s assume a 20% success rate on a retention campaign or plan, which results in monthly rollouts saved as $3,591,000. Now, that’s not a bad number to take to the Board. 

But how do we convert that into revenue saved? Funds under management have fees associated to them, that’s how Superannuation companies make revenue. These can be anywhere from 0.6% to 1.3%, therefore for the purposes of this calculation we will settle on a 1% fee rate. One percent of $3.6M is $35,910 per month, and a not insignificant $430,920 per year.

Unstoppable business case

So that’s how you do it. Feed these numbers into your business case and you will be producing a compelling level of value to your organisation that your CEO and Board will need to acknowledge and support.

Now, all you need is to identify the right partner organisation to help accelerate the application of Predictive Analytics and Machine Learning across your IT and Business teams. 

With significant business challenges impacting the global economy and the superannuation industry, use your data to make a difference. With predictive analytics, your super fund can build optimal retention planning, and proactively address attrition.

Do you need to identify and assess game-changing opportunities, early?
Talk with Tridant to drive rapid implementation of AI to not only predict the impact of internal and external dependencies, but also to gain compelling insights for better business.

Book a Discovery Session with Tridant

Unpack your enterprise challenges. Book a discovery session with Tridant subject matter experts to diagnose, prioritise and create action plans to chart a path forward and solve the challenges you are facing.

Enter your details below and we'll get in touch shortly. 

Recent Articles

Search