In our fifth episode of SME Banking Club Podcast Series, we speak on online lending to SMEs with Bruce Brenkus, Chief Risk Officer at Spotcap.

Spotcap is an online lender to small and medium-sized businesses. The company was founded in 2014 and is led by Founder and CEO Dr. Jens Woloszczak. Spotcap now operates in five countries and has secured more than EUR 65 million in funding. The business has enabled thousands of SMEs to innovate, remain competitive and grow. Spotcap is backed by world-class investors including, Rocket Internet, Access Industries, Holtzbrinck Ventures, Kreos Capital, Finstar Financial Group and Heartland Bank. Spotcap employs more than 100 people globally and is headquartered in Berlin (Germany) with a local presence in Spain, the UK, the Netherlands, Australia and New Zealand.

Olena Gryniuk: During recent years Spotcap entered 2 new markets, which are the UK and New Zealand. How do you estimate competitive landscape there? Especially the UK market is perceived to be as one of the most competitive being a European centre of fintech?

Bruce Brenkus: Interesting question and one we look at very seriously that New Zealand has a very little competition and as you stated the United Kingdom has a varying number of fintech and other players in the market. Both we actually embrace from different spectrums, where we don’t shy away from the competition, we actually embrace it. The competition helps us to that point that the market understands what fintech is, new lending opportunities. Our job is to go there and do it as good as we can, providing the right product and services to the right individuals. So, we can succeed by giving both business and gaining a lot of recognition at the market. And the other market, the other end of the spectrum, the other side of the world – New Zealand, there really is not much competition if any right now. So, our job is to do a lot of the market presence and promotion to get the small business enterprises to understand what are out there and available options. Very diverse questions and opportunities but we embrace these both with different strategies we have to put in place.

OG: Whom do you consider as your main competitors in each market you are present? Did you enter these markets as banks are not too active or not so quick as you are in the lending process?

BB: We do not see banks to be our competition. We see them more to be more of our strategic partners right now and down the road. Our main focus when we enter the market is on the needs of the small business enterprises. So, we look at the regulatory, the environment, understanding industries, the understanding and knowledge of financial lending and perceive all markets under that guidance.

OG: And how Spotcap is different from other players in these markets?

BB: I think our difference is we blend a lot of analytics and automation with human underwriting from very skilled analysts. Our goal is to find the right customers, the right products with the right pricing with a right durations.

OG: What’s the range of your loans?

BB: We start as low as 10000 in local currency whether that’s euros, pounds, Australian dollars, New Zealand dollars, etc. and we go up to 250000 local currency in each market.

OG: Who is your average customer? How does he look like: age, industry, company’s size?

BB: If we look at averages, we probably get into businesses doing business 5-7 years, making revenues between 1-10 mln a year. Probably the director is in 40th and the main industries we are more focused on retail services, wholesale, similar markets

OG: What is the average number of loans that customer can have with you? Do you practice to disburse more than one active loan to the customer? What is the maximum credit limit a customer can have with you?

BB: We provide a line of credit. So, we put that ownership on how the small business enterprise takes the money up to them. So, we provide a line of credit with a certain commitment period, so an amount of time they can draw funds. We go up to 250 thousand. They may all take 250 thousand as one loan, or they may take it in small components, they can do in a bunch in different ways. We leave it up to them based on their needs.

OG: Coming back to your risk model and lending process. How does it look inside Spotcap? So, it takes maximum 24 hours to make a decision, isn’t it? How it goes from the step of receiving an application from the customer till loan disbursement?

BB: Correct. We provide 24H current around from the point we have all the documentation we require which is called a submitted application. Then it becomes a qualitative and quantitative undertaking, where our risk models as plural with multiple models provide back understanding on decision such as with credit create risk-based pricing, how much of the line of credit should we be giving and provides all the recommendations on final decisions along with conditions on that loan that may need to be cleared. So that is where human underwriter comes in. We use a lot of advanced platforms to do a lot of the pre-work, make pre-decisions that let our very smart qualified underwriters take it from there to look at other aspects that we see, because no model can cover every aspect of a lending opportunity for somebody wanting a hundred thousand plus euros/pounds/dollars. So, it takes to look at a lot of other advents. What we do instead of having everybody looking at many different things the models help them pink point the areas they need to look at, clear and make a final decision on.

OG: So, this was my next question what is the human input into your credit underwriting process?

BB: The human part is to take those components in our platform is set up for them to do different visualizations. So, let’s give you a couple of examples to help you: if we have a very small ticket, say 25 thousand in local currency, the automation in the tools may take it all the way to a final decision, when an underwriter only has to validate and verify, almost like a quality control check. Versus on bigger deals the models will provide output and recommendations but will also provide kind of red light, yellow light, green light, it’ll start putting currencies on things that need to be checked that do not either look right or need to be verified and validated, so the underwriter is given a mission to stay on a specific plan to clear those items, to understand the entire process and application to make a final decision, when the underwriter has the ability to override final decisions based on qualifying information, and we have very sound logic and backup – we call it four eyes – where most of our applications above the delegated authority need to be signed off by a second level authority.

OG: Tell us more about the technical part of the process. How do you utilize machine learning within underwriting and the whole risk assessment process?

BB: We utilize machine learning more from what we call in an R&D and laboratory environment. We have something we entire on Spotcap Labs where we take our decision scientists with a lot of our tools and we are always trying to beat the model and production – number one. Number two – we not always trying to look at building a new model but we are trying to identify new data attributes that can be a combination of attributes from certain of our data streams and data vendors. Number two – we are looking at different data vendors to find out if new data sources have any added value. I love the word, and I hate the word “Big Data” because having a tone of data at your table could slow you down. So, what we try to do is sieve through that and find the data that’s important and minimize the other data, so we are not wasting both our time and the applicant’s time as well if it’s not necessary for the decision. And third – we do a lot of tax mining. We have a lot of attributes and data that comes back in human written words, so there is a lot of key information there we are always building and also creating that.

OG: Nowadays fintech lenders set a high customer-service bar for banks and present new challenges for their risk functions. So, as I understand, you do not require a big package of documents from the customer and long loan applicants. So, do you draw some data from public sources like PayPal transactions, Amazon, etc.? How do you do this in Spotcap?

BB: To clarify we do require certain documents from our clients. And that’s part of our not only credit underwriting but authentication process. So, since we are not in a face-to-face environment to get to know them we are asking them for things like their financial statements, we are getting bank account data from them as well, and we are getting their consent to pull the credit report for the director and also the owner of the business. So, using all those different tools we do a lot of cross-checking and verification: does the bank account inflows and outflows match up with the revenues that they are stating, does the data within look like for the retail customer, so we are putting all that data together. So, we are scraping a lot with our robots and putting that data in formal words as easy and user-friendly on our small business applicants as possible. Then using that information to make formal decision after we know the customer or the KYC (Know-Your-Customer) that we have the right business, the right owners or directors of it who can provide the application and we are very deemed to verify that is a legislative business, has an active business licence and have even looked that all the directors to make sure no problems are in this organization or other organizations are tight to.

OG: Many of these digital and technological innovations can reduce risk costs, and bring a competitive advantage to the lending process. However, they may also expose institutions to unexpected risks like: cybersecurity risk and generally data privacy and protection, more risk for fraud due to no face-to-face communication with the customer, the tendency to minimizing the data application process, model risk – which means Banks’ increasing dependence on business modelling, which requires that risk managers understand and manage model risk better, because the consequences of errors in the model can be extreme..So my question is what are the ways you see of mitigation of all those risks in the digital process?

BB: There are a lot of them. So, let’s start with a front-end, let’s go with a making sure we have the right entity in front of us and the right ownership. So, we look at different sources but we do a lot of cross-validation, where the data when the company says they are making 500 thousand this quarter we cross-validate by looking inside their bank account to see if we see the similar inflows and those inflows being from we would call qualified customers, so if it’s retailer there will be all small transactions or they roll-up on a daily basis a credit card. If it’s a wholesaler we’d see something between 3-20 different customers requesting product in certain chunks, so we’d look for similar streams and one of the thing we also do is benchmark that against other companies that we’ve seen in the same industries to look for trends and difference that are anomalies from the normal and if we do those will be kicked out for underwriter to clear and if they have any issues we can push those to the fraud team. Vice versa we have skilled underwriters and underwriters have a tendency when the see things that just do not feel right, and they can see certain reasons they will be able to push that to our fraud team as well to do an investigation. Again, we do that within 24 hours, but we must fully vat the customer to make sure we have the right customer in front of us, that there is no misrepresentation going on and the information provided is solid so that we can provide a sound risk mitigation based decision.

OG: What is loan approval rate in Spotcap? Is it growing? What is your targeted one?

BB: We look for approving around half of the applications, but it depends on the fundals and what streams. We do a lot of business through partnerships which is a lot more controlled and constrained where our partners and customers are trained for what we look for versus when you look at things from searchings and optimizations in marketing would be different. Our main goal when we look really at approval rates is that we are approving profitable customers that have a long gevity, a long gevity could be a long gevity of that line of credit but we also looking for the customers we want to continue to grow with us over time.

OG: What is your delinquency rate, and how do you manage them?

BB: Our delinquency rates are very much within our target range we’ve set up. How we manage it is we have our proprietary risk technology in our models that are the foundations. So, they are Basel II developed from the methodology from probability fall. So what we do is as we build our models and we constantly fine tuning them and we also constantly validate them against real life performance is rank ordering people by what we deem the probability of risk, those models rank order are tight together with risk-based pricing to make sure we are pricing at the right price point and as well our line of credit models or how much we will provide to the small business is tight to models that will make sure that we are setting them with funds at the right duration that they can pay back without any heart check.

OG: Bruce, what is your everyday job looks like? What takes your maximum involvement at the moment?

BB: Everything! My day starts early in Berlin since Australia and New Zealand are in the other side of the world and in a different time zone that based being done, so I get involved with some of the final decisions on larger tickets and communication with the team then the European day starts. So, a lot of interactions with the underwriters to make sure learning from it, the underwriters make final decisions I will sign off on bigger deals. But most of it is really getting involved with the items around, the risk management daily activities, how the portfolio is performing, what’s going on, what’s working, what’s not working, analysing that, working with the decision science and the modelling team in trying to find new techniques, developing models, the normal strategic things to from pricing, investor items we have a lot of investors and what I call “platform evolution”, so we are always trying to make our models work more efficiently with underwriters by providing the right points back and visualization so the underwriter can make quicker, smarter and more informed decisions.

OG: What are your plans for 2017?

BB: Pretty straightforward. Smart, profitable growth. Smart, profitable learning in each of our countries. Our goal is to put into action to keep the business reporting on the book to be profitable – number one. Number two – we are always interrogating and trying to find better ways to do that. So it’s a continuous art.

OG: Bruce, what is your personal source of inspiration for doing a better job? Which country do you see as an example or maybe some particular companies and their models are inspiring you?

BB: First would be – I love the phrase “Can I continuous in never-ending improvement” I’ve been using it for years. It’s important because in this industry you always have to continue learning because the environment and small businesses are changing, economics changes, you always have to be looking forward and as a risk person you have to be very balanced to understand: it’s easy to turn risk off and get zero losses that mean you are putting zero on the books, and that doesn’t work. So my philosophy is always finding the right balance between profitability. So once you put a model on a play we celebrate as a team, we high five, we cheer than five minutes later I’m trying to get the team to be the model that’s are regarded to play. We do a lot of advanced analytics. The philosophy I’ve always had is to put things into actions, so to learn and not use as to not really learn at the first place. So we do try new things and as we are looking at different technologies and things such as Random Forests, Deep Neurals, a lot of different techniques, we wanna make sure that we don’t end up in what I call “analysis – paralysis”, we find things that make sense and put them in action and if there are certain things that don’t work we move forward and look at other things.

The companies that I’d be looking at and my teams look at because that is important to look at technologies in our industry and others. The ones that are the most fascinating for us that we observe are really in other industries that are using the technologies in decision sciences in different ways, so medical research do a lot of in anomaly detection is one of we do a lot of looking at, century type of models that is being used in drivers’ cars like Uber and Google, etc. are ones we are looking at to understand, and then face recognition – the huge one we’re seeing in security and identification. Those may not be used directly today in our industry but we know that they in some manner down the road. Our job is to look at that and to find out how and see if our team can find some of those earlier passes to success.

OG: Bruce, thanks a lot for that conversation!


Please enter your comment!
Please enter your name here