Going Beyond the Score

Abstract: In this panel on the future of credit decisioning, Kevin Chan (CPO, TransUnion) moderated a discussion with Ardalan Shojaei (Chief Credit & Risk Officer, Fig), Becca Mintz (Managing Vice President, Capital One), and John Crum (VP Risk Management, Fairstone) on how lenders are moving beyond traditional credit scores to make better, fairer, and more dynamic decisions. The panel explored the role of cash flow underwriting, buy now pay later data, alternative digital signals, fraud prevention, identity verification, AI, and talent in building the next generation of credit risk models. Across fintech, bank, bureau, and lender perspectives, the message was clear: credit scores remain foundational, but the future of credit will require richer data, stronger identity tools, better consumer education, and deeper industry collaboration to expand access while protecting trust in the system.

👉 Check out the full VIDEO here.

Kevin Chan: Let’s start with some deeper introductions. I want each of you to share your role in your organization, how credit scores are relevant within your role, and, if you are willing, what your current credit score is and one lifestyle habit you want to change to improve it.

I will start by saying my TransUnion credit score that I see in my bank app is an 858. And if I can only kick this coffee habit, I think I could get it higher.

Ardalan, why don’t we start with you?

Ardalan Shojaei: Hello everyone. Great to meet you. I am responsible for overall lending at Fig Financial. For those who might not know us, we do digital lending. We are a fast-growing fintech.

When you ask me about score, we do not have the incumbent status of having relationships with 17 or 20 million Canadians. So we certainly have to leverage the credit risk score. We have been in market for three years, and along the way we have built our own proprietary risk scores. But we also have to do a lot more.

We have to go beyond the score because we have to make really great decisions and treat customers well in order to compete.

In terms of my own score, in my head it was a lot larger than what I saw when I last checked on the weekend. Historically, it was 840, but last I checked it was actually 760. So hopefully there is some sympathy in this group of credit risk professionals for a credit risk professional having a score like that.

Becca Mintz: Hi everyone. My name is Becca Mintz and I lead credit and data at Capital One. As you can imagine, leading credit and data, credit scores and data are massively important to my role in helping Canadians succeed with credit.

My current credit score is 839, which I was feeling pretty good about. I have recently entered a new phase of life. I recently moved into a new neighbourhood, I have two kids, and in terms of healthy credit habits, I am mostly just trying not to screw up this phase of my life. I am trying to stay on top of our household budget and plan ahead for the kids. So for me, I am in a holding pattern right now.

John Crum: Hi everyone. My name is John Crum. I manage credit risk strategy for Fairstone Bank’s Walmart Mastercard and Best Buy Canada financing portfolios.

Credit scores are the fundamental tool that we use in all of our credit strategies to assess probability of default. For our bank, we use a combination of FICO industry scores, and we have also built in-house custom models as well.

My personal credit score, I have not checked in a while. My bad habit is that I am always applying for my competitors’ cards to see the experience. So if you do not care about getting hard inquiries on your file, go for it. But mine has been dinged up because of that.

Kevin Chan: We have all been in the industry for a while, and from my experience, we have been going beyond the score for a long time. We used to look at things like acquisition channel or whether a customer looked at the disclosures. Those were indicators of risk that I have seen in the past.

What are some examples of how you are going beyond the score today, and what is on your wish list for new data sources you would like to see? John, why don’t we start with you?

John Crum: I have two examples.

The first is cash flow underwriting. If anyone does not know what cash flow data is, it is a process where, during the credit application, a customer logs in to their bank account and supplements their credit application with bank account data, with the hope that they have a better chance of approval.

Today, we are slowly testing into this. It is not yet a mainstream process, but based on our analysis, we are seeing some really good lift in the models, particularly in the non-prime and new-to-credit space.

For a customer who does not have a very thick credit file, this is one way they can bolster their chances of getting approved.

Admittedly, the technology in Canada right now is a little clunky. It uses screen-scrape technology, it is not particularly sophisticated, and there can be some latency as the customer waits for the data to process. But in our tests so far, we are finding that up to 80% of our customers are completing this when prompted. That is encouraging. Customers are getting more comfortable with it, especially if they are on the margins of credit and can see the benefit.

The second example is the integration of buy now pay later data into the credit file. A lot of customers using buy now pay later are on the margins of credit. They may be younger customers or customers who have not built a large credit file, and today they are not being rewarded for good payments.

When that data is integrated into the credit file, we think it can help expand access to credit for those marginal customers.

Kevin Chan: Ardalan, anything to add there?

Ardalan Shojaei: Being a uniquely 100% digital lender, I think back to my days at the bank where we would have very different risk profiles for someone who went into a branch to get a loan versus someone who applied digitally online, even if they had somewhat the same credit risk score.

For us, one of our goals is to flip that on its head, where digital can seem less risky because you have so much more data. It is an ongoing journey, because there is still much more to learn and optimize.

There are markets, including in Asia, where lenders look at things like the battery level of your phone, whether you are on iOS, what version you are using, and whether it is out of date. Can some of these signals be combined together and become indicative?

A big part of this is that there is a lot of market out there that is underserved. For a while, people talked about newcomers and using international credit bureau data, but there are also other ways we can get signals so we can serve those customers well.

For us, it is also important to price them well. It is not only the approve-or-decline decision. As a personal loan lender, it is also the amount that you give and the pricing. We want to get that right so we can offer the best pricing we can.

Kevin Chan: I want to double click into cash flow a little bit. Cash flow is something we are looking to try to solve at TransUnion. I wonder if there is a bit of an arrival fallacy here. We all think we want it, but do we really believe we are going to get a lot of value from it?

If we know a consumer’s income and we know their main debt obligations, does all the cash flow information in the middle really give a lot of incremental benefit, or does it create noise in the system?

Becca Mintz: I will take the over. I think it adds a tremendous amount of value, particularly if you are outside the primary banking relationship.

Canada has a unique banking structure with tight walls around some big players. Those players may not see as much value in it because they have access to their own in-file cash flow data. But when I think about some of the customers Capital One serves, including people who are not well served by those institutions, the premise of being able to collect more data to improve underwriting is invaluable.

Ardalan Shojaei: There is certain visibility that we do not necessarily get through more traditional avenues. Even the speed of a new loan being booked can matter. It might take a month or so before it gets reported into the credit bureau. Buy now pay later is also not necessarily captured well today.

If you look at more mature markets, the UK is a good example. They have had open banking for a long time and have been leveraging cash flow, though adoption is still not as high as I would have expected.

The technology adoption is not easy. You do not just turn on open banking and suddenly everyone can start using it. Even classifying transactions is not the same across banks. Is this an e-transfer to a friend? Is this a mortgage payment? There is a lot of clean-up to do.

AI can help with that. Natural language processing and large language models can help classify and synthesize that information, but there is still a lot of work to be done.

Kevin Chan: We talked a little bit about buy now pay later. At TransUnion, we are building something to bring more visibility into it. But when we think about these small microtransactions, do we think there is a huge lift in value, or does it create more noise given that the products are smaller, lower-value loans that may be paid off quickly?

John Crum: I do not know exactly what the overall strength will be, but I am excited to see the data and test it. It will have different strength for different customers.

With buy now pay later, most customers are likely paying it off during the terms. They are demonstrating good transacting behaviour, but they are not getting rewarded for it. To me, that is almost a detriment to the consumer.

Kevin Chan: Switching gears, something we are seeing at TransUnion is a growing intersection between fraud and credit risk. We are seeing more delinquencies that we think are actually synthetic or first-party fraud.

Are any of you experiencing something similar, and what beyond-the-score strategies are you using to mitigate it?

Ardalan Shojaei: As we were investigating some of the insolvencies and bankruptcies on our books, we realized that in some cases the descriptions said the customer had been the victim of an investment scam.

It is quite sad. Fraudsters go to the point of least resistance. As institutions have invested in frontline defences to prevent account takeover and hacking, we have seen fraudsters and international scam organizations move toward victims who have legitimate credit scores and are doing quite well.

Suddenly, that person voluntarily gets a loan and hands it over, along with their life savings, to a scam organization that they believe is a legitimate investor.

Cash flow has helped us a lot here. We looked at transactions and identified when an individual was sending funds to scam organizations. We have built a negative database, and I am happy to share that with others because we all want to stop the bad guys from doing their job.

Where we do not know, we are using AI agents to try to develop a trust score through digital means. And we also make an effort to reach out to the customer. Even if we do not give them the loan because we know it is for a bad purpose, we will warn them about it.

There is also a Canadian anti-fraud coalition now, and I encourage people to check it out and share it with others.

Becca Mintz: I would add another customer angle.

In addition to people getting scammed, fraudsters often target thin files. They may create a synthetic identity or take over someone who does not have a rich credit history. That is such a vulnerable part of our country and our ecosystem.

It also inflates the losses in those segments. If you are looking at a marginal risk segment and suddenly you have higher losses, a big chunk of that might be synthetic or first-party fraud. That can make less sophisticated lenders less willing to grant credit to those segments.

I am fortunate to work at a company that has best-in-class modelling, and this is a huge focus of what we do when building credit risk models. But for competitors trying to enter, it would be really hard to play in a space where the convoluted nature of fraud and credit risk is so muddy.

John Crum: I remember when I first started in credit cards and chip and PIN had not been widely adopted. The big source of fraud was counterfeit fraud. Scores were developed to identify it, and fraud teams were kept extremely busy.

The response was that chip and PIN was adopted in the payments industry, using a dynamic cryptogram to communicate between the card and the terminal. Since then, counterfeit fraud has essentially been eliminated.

I think card-not-present fraud may go the same way. As more purchases are digital and authenticated through a device, fraud is going to move almost totally into the identity space.

The way to combat that is to take lessons from chip and PIN and create something like an identity token. Today, we can use scores and models, and we have technology for ID verification, but it is not perfect.

The more perfect solution would be an encrypted token. Maybe in the future, when I am communicating with you, I am sharing my encrypted identity token. But building that will take time and require government involvement.

Kevin Chan: If we were better at identity verification, do you think that actually opens up more access to credit, or are we just going to be better at saying no?

Becca Mintz: I think both.

When fraud and credit risk get blurred, the problem is that lenders struggle to split the good from the bad. In those cases, you might say no to a lot of good people, or yes to a lot of bad actors.

Any data that allows us to build models that better rank order the likelihood of fraud or default should allow us to identify more people who are on the upswing from a credit perspective and carve out bad actors more efficiently.

Ardalan Shojaei: We are also talking about something like universal ID. If you think about it, 20-plus years ago, the amount of SIN captured by bureaus was around 98% or higher. Now it is around 33%, and it keeps going down because it is optional.

As we think about coalitions around scams, you want to identify whether someone is a fraudster. But it is tough to make the match. We are worried about privacy, and today we often have to share name, address, date of birth, email, and other details.

A better identity system could help the industry, and in some ways it could also protect customers from having to share so much other information.

Kevin Chan: We have talked a lot about data. But beyond the score can mean much more than data in the credit landscape, especially as we see changes like AI, real-time interaction channels, and increased privacy regulation.

What are the non-data elements in the credit ecosystem that help us go beyond the score?

Becca Mintz: Non-data, Kevin? This is a room of data people.

I am going to go with talent. It is probably the most important thing for any of us and the companies represented in this room.

As our jobs evolve over time, I spend a good chunk of my time thinking about bringing the right talent into Capital One. Ideally, we have people who are passionate about credit, data, and modelling, but the talent model has completely shifted.

The type of work I was doing in my early days at Capital One does not exist in the same way anymore. The coding languages are obsolete, and the repetitive tasks and basic data hygiene work are completely different from what new analysts are doing today.

We are asking more junior people to have critical thinking skills and the ability to effectively challenge work almost right away, especially if they are going to rely on AI agents. We are at a massive inflection point in talent.

John Crum: I totally agree. The skills that have always been important in credit decisioning and credit risk—curiosity and creativity—will still be needed, and they are more important than ever.

We have to figure out how to use AI to make our businesses better. Does that mean improving productivity? Improving credit decisioning? How do we leverage it in a safe way?

AI will happen no matter what. If you do not adopt it, you are going to get left behind. You have to keep your mind open and get creative about ways to use it.

Ardalan Shojaei: The creativity part is really important.

Not long ago, people looked at credit systems almost like black boxes. Some credit analysts lost curiosity because they knew what went in and what answer came out, but they did not understand the mechanics of what the model was actually doing.

AI is making things so accessible and easy that there is a risk of short-circuiting the thought process and the experience that builds creativity.

Externally, we also need to think about how we engage with customers differently. Today, the experience can be very rigid: question one, then question two, then three and four. For the right person, maybe you need different flows that are more dynamic and specific to the individual. AI can help enable that.

Kevin Chan: We have talked about data, policies, AI, and talent. If there was only one thing you could change right now about the credit ecosystem, what would it be?

Ardalan Shojaei: I am really passionate about the scam and fraud aspect because we lose so much money, and unfortunately vulnerable people are impacted. I would love for the industry to come together to fight scams even more collaboratively.

Becca Mintz: I will take the consumer angle. I would like to see more education around how credit works and how credit scores work.

If you have ever spoken to a 17-year-old who is about to turn 18 and is on the cusp of being eligible for credit, and asked them how credit works, it is bleak out there. I would love to live in a world where we value education on how credit works more than we do today.

John Crum: I agree with Ardalan.

Everything else we talked about today, whether it is cash flow, buy now pay later, or AI, can be considered competitive initiatives. Each bank may have its own approach, and it may or may not fit within that bank’s risk appetite.

But one thing we can all agree on is making sure customers have trust in the banking system in Canada. If we do not tackle fraud issues in a coordinated way, we are going to erode trust. That is something we can all agree on.

Kevin Chan: Thank you so much. Thank you for joining our panel today. The future of credit continues to be super exciting. We at TransUnion and with the CLA are excited to partner with you in this future.

Here are 10 key insights from the panel:

1. Credit scores remain foundational, but they are no longer enough
Panelists agreed that scores still play a central role in underwriting, but lenders increasingly need additional data and models to make better decisions.

2. Cash flow underwriting can help expand access to credit
Bank account data can provide useful lift, especially for non-prime and new-to-credit customers who may not have thick credit files.

3. Customer adoption of cash flow tools is improving
Fairstone’s testing found that up to 80% of prompted customers completed the cash flow process, suggesting consumers may be willing to share data when the benefit is clear.

4. Buy now pay later data could help reward good repayment behaviour
Panelists noted that many consumers successfully repay BNPL obligations, but that behaviour is not always reflected in their credit files today.

5. Digital lenders can use alternative signals to understand risk differently
For digital-first lenders, behavioural and device-based signals may help create more dynamic risk assessments, especially for underserved segments.

6. Fraud and credit risk are increasingly connected
The panel highlighted how scams, synthetic identities, and first-party fraud can appear as credit losses, making it harder to distinguish bad actors from higher-risk borrowers.

7. Better identity verification could increase access, not just reduce approvals
Stronger identity tools can help lenders say no to bad actors while saying yes to more legitimate consumers who might otherwise be rejected.

8. Industry collaboration is essential to fighting scams and fraud
Panelists emphasized that fraud is not simply a competitive issue. Coordinated action is needed to protect consumers and maintain trust in Canada’s banking system.

9. Talent and critical thinking matter as much as data
As AI changes credit risk work, organizations need people who are curious, creative, and able to challenge outputs rather than simply rely on black-box systems.

10. Consumer credit education remains a major gap
The panel closed by noting that many young consumers do not understand how credit works, making education a key part of building a healthier credit ecosystem.

Sign up for the CLA Finance Summit Series