The secret to customer behaviour

When Alan Ringvald and Abraham Rodriguez co-wrote their Masters thesis on the science of unresponsive customers, they did not predict that it would lead to a machine-learning platform that can accurately forecast when customers are about to jump ship and when they are ready to repurchase.

The two MIT students had approached a number of large companies for anonymised customer data sets to assist in their research. They were surprised when not only were the corporates willing to share the data, they also offered to pay them to crunch it.

‘The intent wasn't to start a company out of the thesis,’ says Ringvald. ‘We approached these companies in the spirit of search but, within the final few days of school, we decided to try it out as a business.’

The company they launched in 2016 is Relativity6, which arms insurers and brokers with the intelligence to predict customer behaviour via its proprietary behavioural economics algorithms.

It takes anonymous customer data and analyses their past purchasing behaviours. It can then predict which lapsed customers will repurchase and which product or service they are most likely to choose. It can also predict how a customer is likely to re-engage, such as by phone or email.

Coupled with its cross-sell recommendation algorithm, Relativity6 can reduce monthly churn rates by an average of 1 per cent. Its financial services clients include Allianz, Chubb, Citibank and Willis Towers Watson.

INSIDE THE ALGORITHMS

Ringvald will be among the speakers at this year’s ANZIIF Insurtech Conference, where he will share insights into the successful partnership between Relativity6 and the Melbourne office of Willis Towers Watson (WTW).

The partnership was formed in 2018 and WTW was immediately able to predict which customers were likely to leave with 84 per cent accuracy, 60 days before they actually left.  

Within the first quarter of Relativity6 weekly predictions, WTW achieved an 80 per cent effectiveness in retaining clients identified as being of high risk of churn.

‘I’m going to take the audience through the full story, including how we build a product that works for the humans involved with the prediction,’ says Ringvald.

FINDING MEANING IN THE MACHINES

Beyond the academic world, Ringvald had worked for Google as Partnerships and Account Executive Product Marketing. He had also co-founded start-up companies, including sports analytics business Superdigital.

‘Coming from industry, Abraham and I  both realised that just because machine learning is valuable, it was very difficult operationalise those algorithms to create value for a business,” he says. ‘I think it's still the case that data science in general is more experimental, scientific, and not as practical as it could be.

‘What we were thinking about in school was how we could work with a large enterprise to help them create useful machine learning models that actually made sense for business purposes.’

FROM IDEA TO ENTERPRISE

The co-founders initially raised US$700,000 to enable them to hire a data science team to service the customers they had gained while completing their thesis. To date, Ringvald says they have raised about US$1.6 million.

‘We're effectively profitable, which is really nice,’ he says. ‘Eventually, we're going to need to grow and maybe take more money, but the primary goal from the beginning was to create a company with a real foundation, where we were making more money than we were spending and growing at a reasonable rate with our customers' help. If they were happy, they would refer us.’

PREDICTING CUSTOMER BEHAVIOUR

Relativity6’s platform identifies the windows of time where people's behaviour has a propensity to change.

‘We have the ability to tell our customers which clients they should speak with this week because they might leave or might want to buy something else,’ says Ringvald. ‘The reality is that our lives are dynamic, so how much we might spend with a company is going to change constantly. The problem is that most companies create one calculation to say, “This is the value that we think this customer will bring,” and that's it.

‘Our technology is constantly recalibrating in time,’ adds Ringvald. ‘That's also why the company is called Relativity, because we know how important context is for valuing someone.’

THINKING OF SHOPPING?

Relativity6 can also help companies to understand when a customer is merely thinking of shopping.

‘We are there before they're even on Google looking for an alternative,’ says Ringvald. ‘We can be two to three months early before that shopping behaviour exists, which has been an interesting differentiator for us.’

Another point of difference for Relativity6 is that its platform is built with humans in mind, says Ringvald. Its aim is to creates models and experiences that help company drive greater value.

‘The predictions that we're creating have to be used by the people who are making phone calls or writing emails to their customers. So, our technology is extremely understandable.

'It's easy for a human to interpret why we think someone's going to be lapsing, or why we think somebody might want an alternative product.'

DATA BASED ON INTUITION

Relativity6 has worked with WTW to build a product that creates new data based on the intuition of its brokers.

‘It's an interesting human-machine interaction,’ says Ringvald, adding that this will be part of his presentation at this year’s Insurtech Conference.

‘At the end of the day, we're a machine learning company. We're not the ones talking to the customer every day.

'We realised it was really important to observe the behaviour of the people who are closest to the client and use that as a nominal data point and put that to the prediction. This human-machine interaction was also incredibly important in making our technology better.’

Relativity6 employs 25 people who Ringvald says are ‘hired to do their part’. The company has clients so Asia, Europe, Latin America, the US and Australia.

‘Our technology is agnostic for language, we can work in all kinds of different use cases,’ he explains.

‘We love working with insurers and  large enterprises. Our plan is to expand the company that we have built within the financial services industry. We are focused on exploring other ways that this technology can be applied within the insurance space.’

HANDS-OFF LEADERSHIP

Ringvald describes his leadership approach as ‘pretty hands off’.

‘My philosophy is you learn from actions, as opposed to constantly having meetings and talking about stuff. I'd rather keep things in action, get feedback from that and then make decisions.

Ringvald says this can lead to mistakes. ‘It means we're willing to fail a lot,’ he says. ‘It may create a more painful experience in the short run but, ultimately, it creates a more valuable organisation because you're not just going off your own gut. It's based on what we're seeing the customers do and what they want.

‘We've built a lot of products based on customer feedback rather than us being in a room and deciding what we think is best for the customer,’ adds Ringvald. ‘It seems to be working.’

Hear Alan Ringvald ispeak at the 2020 ANZIIF Insurtech Conference. Register now.