Buying a property to live in is often the biggest financial commitment a person will make. Financial technology company Trussle aims to make this process easier by matching consumers to the ideal mortgage or remortgage product to finance the property through a recommendation engine. Robert Fairfield, the VP of platform, and his team learnt four big lessons during and after creating the engine.
The recommendation engine works by taking 3,000 to 4,000 discrete data points from each customer. These data points will be hard or soft. The hard ones are about the person’s age, income, location and things that are easy to quantify.
The soft data points are about the customer’s preferences as well as their attitude to risk, and the likelihood of their circumstances changing such as an increase or decrease in income or starting a family. The engine also takes into account facts about the UK housing market, on a macro and a micro level.
Fairfield said: “We take all that information and mash it all up to try and end up with a single recommendation for a customer and say ’of all of that uncertainty and all of that information, this is the right product for you’.”
"They know they are going to sell, they are built on demand."
The most important lesson he and his team learnt, after ironing out some of the bumps, is that there will always be some people for whom a product cannot be found. In that case, build a product for them. This, for him, is an opportunity. He said: “That is incredibly exciting. The next wave of product innovation is going to come from those those companies that have identified that and gone out and built specific products for those customers. They know they are going to sell because they are almost being built on demand.”
The customers that are currently underserved by the mortgage market are the self-employed and the older generation. “We know they aren’t getting a great deal so we are building products with lenders to service that.”
He added that more and more technology companies will start to do this and gave the example of a tech giant that has followed this model of product development. “When Netflix create content, they know it is going to be consumed because they are identifying areas where the recommendation engine comes up short.”
Another important lesson learnt on this journey is to be transparent. “Don’t build a black box,” he said. Fairfield saw that in the past, the recommendation engine was working well and Trussle was presenting mortgage products to customers with a high degree of confidence that it was the right one for them. But the customers often would not take it. “We realised that, especially for quite complicated decisions with serious consequences like a mortgage, people wouldn’t just believe us and take our recommendation.”
This was because it was not clear how the ultimate decision had been reached. Mortgages can be fairly complex. Some mortgage products with higher interest rates and lower fees can end up being cheaper that superficially more attractive products that have lower interest rates and higher fees. When Fairfield and his team started to explain the decisions made by the recommendation engine to the customer, they started to get a much better rate of conversion.
"To buld trust, educate and empower your customer."
He said: “We began to show the next best alternative and explain why that was not recommended to the customer. Trust is incredibly important; it takes a lot of time to build. The best way to do that is to educate and empower your customers.”
The third lesson is to start small with data collection. Fairfield said that at the beginning he gave in to the temptation of collecting as much data as possible. Then trying to assimilate and store it and get some smart people to look at it to find interesting insights. “We wasted a lot of time and energy doing that. We realised very quickly that all this data was in the completely wrong format. We didn’t even know what some of it was,” he said.
His advice is to smart collecting data on a small subsection of customers and on a small subsection of the product set. In addition, be very purposeful about what data to collect about the customers and what value you want to deliver to them before iterating on it. “Only when you start to see improvements should you really build upon it,” said Fairfield.
His final lesson is the biggest benefit might not be in the form that you expect. Fairfield said: “By deeply understanding our customers and what they want, we could change the journeys that they go through for a far bigger impact.”
He and his team soon realised that customers looking for mortgage products can quickly be divided into two groups; those looking for mortgages and those looking for a remortgage. The person getting the mortgage wants a lot of support and explanations about this type of product, whereas the person who wants the remortgage wants to answer the questions, execute and move on.
As a result Trussle created a different digital journey for both types of customer. “We’re using the information we’ve learnt from our decision engine to think about, not only what do we provide our customers as products but how we provide them, when we provide, at what point we offer them customer service, what proactive content we provide throughout the journey.” Fairfield said that this has had an effect on the conversion rate, the cost of sale, as well as improving the final decision itself.
So by starting small with data collection and looking out for what some customers wanted but couldn’t get, this company was able to tailor customer journeys and help create bespoke products for a receptive audience.