In a recent webinar, I caught up with OppLoans’ Matt Gomes on life, loans, and analytics.
Matt is Senior Director at OppLoans — a Fintech company hyper-focused on increasing credit access for middle income credit-challenged consumers — folks with a <600 FICO score. (Think “rescue-rehab mission” but with loans.) With OppLoans, clients graduate up the FICO spectrum and gain access to traditional forms of financing.
From iterating on the number of steps in their application process to eliminating drop-downs and using more radio buttons instead, Matt shares what OppLoans did to rev up their Conversion Rate Optimization (CRO) engine.
You can watch the full conversation here, or grab the highlights below.
1. Can you talk about how important your website is to your business and the industry?
Our website is the lifeblood of the business. 100% of our applications are sourced online. We made the pivot to fully digital about five years ago. I think digital is becoming more and more important for traditional financial institutions.
I do 80 to 90% of my banking on my cell phone or a laptop, and that shift is only going to get stronger. Especially in a pandemic, people don’t want to go stand in line at the bank.
2. How did tracking users evolve as you moved to 100% online?
When I joined, we had a very janky setup with Google Analytics with someone else contracted to set up the tags and all the tracking. Pretty quickly, we figured out it was just outdated data. We didn’t have the ability to update the tags ourselves and what we thought were quality applications, weren’t. It was a bit of a disaster.
We knew the application count. We had the credit performance. We had the loan data. But we really didn’t have a good sense for further up the funnel.
Back in 2016, we had probably five engineers total and we really didn’t have the bandwidth to dedicate one or two people to tagging the website. And my team was just me, splitting my time between analytics and a marketing channel.
What really attracted me to Heap is we dropped in the snippet and immediately started tracking, and we proved ROI in just 30 days of data. Now we use Heap to track 100% of what happens on the front end, and a decent amount of what happens on the server side. Combining Heap with Contentsquare and Snowflake, we have a 360-degree view of every session.
Improving conversion events by targeting micro-conversions
3. How do you report out all the different tests and the conversion funnels that you care most about?
When testing something in the application, we want to carry that analysis all the way through.
How do they perform at that first choke point (which is the business rules and initial underwriting)?
Did they get through bank verification?
Are they signing the contract and following it all the way through to a funded loan?
I think one of the really nice things that we’ve been able to do is allow non-technical users on the marketing team, product team, or on the design side to self-service their analyses in Heap’s web interface.
Most commonly a channel manager will come to us with something they put together in Heap and say, “Hey. This doesn’t quite look right. Did we break something?”
That’s one of the things we’re always looking for. Then we know where to look and we’ll dive in under the hood.
4. What are some examples of how tracking all this data has been helpful to your team? Biggest lifts?
What we started to see in 2017 was this slow, gradual decline of folks completing the app. In diving a little deeper, what we were really seeing was flat performance on desktop devices or laptops.
On the mobile side, it was actually pretty substantial month over month over month, which makes sense because there’s been a greater and greater shift to just mobile traffic. At the time, we had a five-step application process…so we thought, what if we made it three? (The final step is always disclosures and disclaimers.)
Hypothesis being, if we’re losing 10-20% each time we make you click forward, we’ll take a bigger hit up front, but we’re going to remove two cliff conversion points. And sure enough, we saw about a ten-plus percent lift in app completion. But that change alone delivered well over seven figures the first year that we rolled it out!
(A 10% lift in app completion that led to over seven figures?! Download the Heap Guide to Conversion Rate Optimization (CRO) for a systematic, repeatable approach to improving conversion through your digital experience.)
The beauty of CRO testing is it’s the gift that keeps on giving. Over time, its snowball effect is incredibly valuable.
The other biggest lift is when we thought: “Now that 90% of our traffic’s mobile, what if we made the application look like an app?” So, we went completely in the opposite direction and made a 16-step application flow. But there’s one question per page, and it’s very easy to just pick the answer.
We’ve eliminated drop-downs and added more radio buttons and graphic icons. That drove a 2% lift on the app conversion — which is still very substantial, close to a seven-figure impact — but what it really did was improve the accuracy of the information we were getting.
So even though we weren’t getting a lot more people to complete the application, we were able to underwrite them much more effectively. Our ability to close the loan and fund it improved by about 10%, which was huge for us.
5. Any recommendations for how someone reading this can improve their website experience?
I think it depends how mature your company is. First and foremost is just making sure things aren’t breaking all the time. That is probably the biggest source of your customers’ frustration. When we first started, we were constantly monitoring conversion from step to step.
Did it suddenly nosedive? If it did, something probably broke.
Did people suddenly stop signing contracts? Could be that they’ve stopped rendering appropriately.
It was less about improving, and more about protecting the core experience so folks didn’t get frustrated. Once you’ve got that foundation, that’s when you can really start testing and iterating and leveraging the data you have.
If you are a really established company, the key is to have a complete view of what your customers are experiencing. We rely heavily on Heap, but we have a fully encompassed data set. It’s very, very hard to get that kind of cohesive data set at a really large, entrenched organization.
I think the more data you have, you can get a couple of bright analysts and data scientists and they can point at the next place you should be investing. Then you’re able to move forward at a much faster pace.
6. As you iterated on your application funnel, what worked well and what could be improved?
We have yet to run a test where we didn’t know what the outcome was, so that’s worked very well. Having to repeat tests because we didn’t have the tracking set up appropriately is a pain point I’ve felt earlier in my career.
What we could improve on is tailoring our application to where each customer comes in from, and making sure that it’s as seamless as possible. This could mean taking away fields because we already have that information. It adds a lot of complexity to our code base, but that’s something we’re going to invest in in the next year or so.
Defining events with the Heap Data Dictionary
7. Are you also using Heap to track retention as well as conversion?
I think a lot of startups forget it’s always cheaper to retain a customer and continue serving them than it is to go and acquire new ones. One of the challenges we face is, once we meet your financial need, and you’re making loan payments — you’re set. You’re not logging in, you’re not checking your account status.
So we’ll re-target folks that have fully paid down a loan with email campaigns to remind them, “Hey! We’re there in the future if you need us.”
The problem is since people don’t routinely log in, they have no idea what their password is. The actual email performance encouraging people to come back, refinance, get a lower rate was great. Open rates were fantastic. Click-through rates were great. But four out of 10 people couldn’t log in and that’s a terrible experience. Even if they click the link, suddenly they’ve lost the motivation.
It’s probably a 35% hit that we’re taking at that login stage, which is certainly substantial. So, we’ve begun testing additional ways to kind of lighten that login process. That’s going to be a really big focus for us. And every single time we test it, we’re following whatever the data is telling us.
8. How reliable is the Heap data when analyzing customer behavior — will it miss any actions, or have you had any problems?
I have yet to see any tool in a tag capture an absolute 100% of data. But I would say we see a less than a 5% gap over any period of time. From an actual data collection standpoint, we have not had issues.
Where we have had problems — and this is more of an internal thing — is in standardizing the approach to creating events, like having the ability to control user permission so we don’t have the Wild West in our event definitions. That’s more about requiring everyone in the firm to follow the same approach.
9. Any thoughts on when to look at data in-depth, and when not to stress too much about it?
Anything we’re testing will go on a maximum of 25% of traffic for the first 24 hours it’s live. Assuming nothing’s broken, we’ll expand that to 100% traffic at a 50/50 split. And it will be run until we have a minimum of 1,000 observations in each variation.
Anything below that, it’s certainly possible to have a statistically significant result, but it’s a really small sample size. I think it’s worth slowing down just enough to get that confidence so you can make better decisions.
(At Heap, we agree that early results can definitely change with the masses. So, our recommendation is to wait until things have settled and you’re starting to see some of the same trends with the change that you’ve made before making any operational changes.)
10. If you didn’t invest in a solution like Heap, what would have happened?
That’s a great question. I don’t think you can actually operate without a solution like this. The investment to do it internally would be astronomical. For the cost of engineers in today’s market, you can probably come up with something better for them to do than purely tag what other people have built.
There you have it — OppLoans’ take on CRO and the impact it’s had on boosting conversion rates and multiplying ROI. Learn more about OppLoans’ story to see how the Fintech firm uses Heap’s autocaptured data to offer customers the best digital experiences possible.
Heap helps financial services companies across banking, insurance, lending, and fintech understand their users and build modern digital experiences. A better digital experience means higher conversion rates, higher share of wallet, and higher customer retention. Visit Heap.io/FinServ to learn more.