AI to help banks retain mortgage customers developed by Elula in Sydney

By developing algorithms that use big data sets generated by banking activity and the property market, the whole company has been built around this single issue of preventing mortgage customers from switching.

Mr Shipman was formerly CBA’s head of automation and robotics operations; Ms Russell was a general manager at CBA for a decade. Former CBA chief executive Ian Narev is helping Elula as a strategic adviser, and is also an early investor (although CBA is not, at present, an Elula client).

However, one of the other major banks is using Elula’s technology, and it is in talks with a second of the big four.

Already, it has signed up 20 lenders, including AMP Bank and Suncorp, some credit unions and mutual banks, and a handful of non-bank lenders.

“Banks that can deliver personalised experiences combined with proactive engagement will win this market,” Mr Shipman said.

Using AI to retain mortgage customers is just the latest application of the machine learning and related technologies by Australian banks, with use cases extending from customer service chatbots, to compliance with “know your client” (KYC) rules and fraud mitigation, to helping banks work out what products to offer customers, and assessing the credit worthiness of their customers.

How it works

Elula’s system helps banks identify which of their customers are most at risk of switching by ingesting around 20,000 attributes that use combinations of customer data held by the bank, including transaction accounts. It also uses macroeconomic data, including about bank competition and pricing, and the national housing market. It does not use any individual data outside what is generated by banking activities with the bank.

One of its products, Sticky, predicts and then ranks customers most likely to churn within three months, either through refinancing with a competitor or selling their property. It then suggests to a lender the right conversation to convince the customer to stay.

Another product, Nudge, predicts which customers are likely to apply for new lending within a six-month period, allowing the bank to proactively serve up a credit offer. Machine learning helps to improve the models over time.

“Elula’s AI technology and ‘explainable AI’ delivers a bank’s front-line staff the next best conversation to have to retain and best support their existing customers,” Ms Russell says.

With more than 1 million Australian refinancing home loans in 2022, Elula says rates of customer churn varies considerably; some lenders have switching rates under 5 per cent, while others are peaking at 24 per cent.

Based on its experience with existing customers, Elula claims to have reduced churn rates by 60 per cent.

It estimates that if its technology was applied by the big four, it could retain between 7500 and 30,000 loans each year, depending on the size of the bank, representing $2.25 billion to $9 billion of home loans. It’s a number that represents significant profits, given mortgages are high returning financial products.

Mr Narev has been joined by Paul Rubenstein, managing partner of law firm Arnold Bloch Leibler’s Sydney office, as an adviser and an early investor in Elula. PEXA, the online conveyancing platform, took a 26.5 per cent stake earlier this year for an undisclosed sum.

Mr Shipman said the principled adoption of artificial intelligence, which maintains customer privacy and adopts strict governance standards, is at the core of Elula’s purpose and strategic thinking to ensure it can exceed banks’ very high bars regarding the ethical deployment of AI systems.

Leave a Comment