Banjo Loans plucks data from the cloud, reduces risk, tunes up for growth

In the last financial year just over 900,000 businesses with fewer than 20 employees were operating across Australia. This is the engine room of the economy.

To grow, those businesses need a great idea, talent and capital. Banjo Loans was established in 2014 to help with the latter, because the founders believed there was a gap in the market not being fully served by the major banks.

Research commissioned by Banjo indicates that in the coming year two thirds of small and medium enterprises will want to borrow funds in order to fuel their growth. Banjo is ready to support them and is now developing a new technology platform that uses intelligent analytics to calculate risk and determine whether a business is a good bet.

Born in the cloud and with its core systems built on Microsoft Azure, Banjo has recently worked with Microsoft partner Data Addiction to scope the new platform which is designed to deliver best-practice data analytics and reporting and lay the foundations for a machine learning AI stack.

Microsoft caught up with CEO Guy Callaghan and co-founder and chief technology officer Julian Hedt to learn more.

Microsoft (MS): Tell us a little bit about the way you are set up?

Guy Callahan (GC): I think of us as a FinTech combined with financial services business. The DNA of the business is that we provide great financial products and services and combined with great relationships. The couple of areas that we really stand out are speed to market – the ability to turn around the applications really fast –  and offering unsecured loans to businesses.

The reason we can turn them around really fast is because of the FinTech side of it – the systems that allow us to really quickly gather the information we need on clients and then make an assessment of their loan application.

MS: What led you to Microsoft as the best platform for Banjo?

Julian Hedt (JH): I came from a history of working in large banks and building technology, and it had always been on Microsoft platforms. We took a bit of a punt that far back, going with Azure, but we’ve never regretted it. It’s been fantastic seeing the continuous development and extra features that Microsoft keeps adding, and us being able to take advantage of them and roll them into our platform for use by our customers.

MS: You’d have some regulatory and reporting requirements too I guess?

JH: We’ve always had fairly sophisticated and complex reporting requirements – regulatory reporting and our funding partners require a really high and detailed level of reporting on how our business is going. The board has always been very interested in seeing numbers to back up where the business is going as well.

Although reporting was about 80 per cent automated, it was complex and there were some manual steps.Our business has grown to a size that, if we keep going the way we are, something’s eventually going to break. If we move on to a new platform, then we’re going to have all of these extra features – things like customer-driven reporting and analysis – and a lot more complex data analysis.

MS: So you worked on this with Data Addiction – are you also developing in house capabilities?

JH: Data Addiction was the partner that we worked with here in Australia, and they were fantastic. They had the internal processes and expertise to stand up the platforms and have an end-to-end solution ready to go in quick time.  Otherwise we might have spent 6-9 months trying to stand up on our own. That set us up with a foundation for what we’re doing now. We’ve actually had some hires recently that are starting to make a lot of use of the platform and extend it with things like machine learning. We now have a data scientist, Scott, on board and he is extending the types of data we’re pumping through it. He’s already added a few machine learning models, has started doing some of our regression analysis over our portfolio, and enhancing our risk engine.

MS: What Microsoft platforms do you use?

JH: We had Data Factory storing our data in Data Lake. Then we used Databricks to do a lot of the transforms. We’ve got Azure Synapse and on the reporting side we’re using Power BI, which is great because we can actually push out web reports to all of our people. And what Scott’s been doing is hooking in Azure Machine Learning services into that pipeline as well.

MS: And what has the impact been on your reporting and analysis?

JH: In our team showcase someone asked a question and said, “Oh, are you able to slice the data a particular way?” And in 20 seconds, Scott had actually pulled up a Date Slider and could actually cut and slice it by date. Now, in the past, that would have been probably a day or two of someone going off and trying to put it into Excel.

GC: I regularly need reports and forecast data for the Board, investors and strategic partners. The ability to access the data immediately in a way we require presents a massive upside.

MS: This new data platform extends beyond reporting doesn’t it?

JH:   Banjo is data-driven. We pull in financial information from 10 or 15 different data sources. We’ve got our own internal credit engine that assesses loans based on business performance. So we’ve always used a lot of data and a lot of heavy analysis. What’s exciting about the data platform that we’ve got now is we can start tuning our products for the market a lot better.

When we first started, we had a very simple product. It was a six month loan, up to $250,000 business working capital. Now we have four or five different types of product in a much broader range.  As competition has increased in the online business lending space, products are becoming targeted towards specific customer niches.

We’re becoming a lot more sophisticated in how we roll out products, how we target them to particular market segments, the information that we ask from the customer to assess a product.

Let’s say you want to borrow $50,000. A lot of customers expect to only provide the minimum of data for that, and expect our decision to be super quick. Whereas if you’re borrowing, say,  half a million dollars, we can ask for a lot more data from the customer. The platform will help do the analysis to say, “For this particular financial product and this loan amount, this level of data coverage is needed to be able to make an informed decision about the risk.” So it’s really helping us fine tune our product suite, and our risk assessment.

MS: So this is going to transform Banjo?

JH: It’s almost limitless – we do a lot of financial analysis about their business that customers would probably find useful, and that could actually help them to run it. So there’s potential there to provide dashboards back to customers with analysis of their own business that will really help them.

MS: Will you also harness AI?

JH: AI is really good for specific problems and pattern recognition. Maybe most importantly, it can sometimes discover related factors and patterns in our data that conventional credit risk analysis may not. What I’m keen to see with AI is the credit risk model optimisation, and if there’s ways that we can improve our analysis and our risk assessment of customers. So for specific products there may be some factors that are much more relevant than others – maybe it’s which city or suburb you’re in, maybe it’s the business experience of the applicants, maybe it’s the type of industry that the business is in, that may really have a strong correlation with good customers.

The flip side of that is to find out if we could be doing business with customers and lending to customers that we’ve perhaps knocked back because we’ve looked at them in a historical banking manner. Machine learning could help us identify customers we could have, that we currently don’t.

MS: Clearly a lot of the data that you hold is private and confidential – how do you secure it?

JH: This is one of the reasons we chose Azure at the very start. Azure Active Directory allowed us to integrate security across all of our Azure stack, and across Office 365. We’ve got role-based permissions, users and groups, and two-factor authentication as a core part of our platform.

The security dashboard in Azure is really fantastic. It keeps you honest, because it keeps coming up with recommendations and you always feel a bit guilty if you let your numbers slip back down. That was probably the intent. So thank you Microsoft, you’ve made every CTO’s job just that little bit trickier. But joking apart, having that single view of all of our infrastructure and our data, and are we following security best practices? I don’t know how we could do it manually without a huge security team.

MS: Looking forward, what does this new platform promise?

GC: Our whole purpose is try and empower growth for our clients. We want to use our smarts and our technology and our partnerships to help them get to the next stage of their growth. “Smart Solutions” is one of the key pillars we live by at Banjo.