Use Data & the World Is Your Oyster

Omri Yacubovich
7 min


As banks face the challenges of an impending recession, it’s time to shift toward data-driven decision-making and AI integration. It all starts with having the right strategy, the right tactics, and the right mindset — none of which have anything to do with the bank’s size.

Dispelling fears of job displacement, Chris Nichols shares with us how he envisions a future where AI and human expertise work in harmony, boosting efficiency and customer satisfaction, and hold the key to a thriving small business lending market.

“It is, in fact, the new oil […] We can help the customer so much better if we know what they’re doing and have that 360-degree view. But it all stems from having the right strategy.”


Omri Yacubovich: Chris, let’s start with the current lending landscape. What’s the best path for business owners to find reasonable terms, at a time when many bank lenders are just pulling back on offering credit?

Chris Nichols: I think the whole industry is pulling back. The first quarter hit 2008 levels, which was a little shocking. With the rise in rates, we have seen some slowdown in the economy, as the Federal Reserve and most of the market predicted. What I think is not predicted is the extent and timing of the recession.

Almost every credit decision that banks make is done with that lens of a recession on the horizon, and I think it’s time for banks to change their lending strategy. The higher probability of a recession means going after higher-credit-quality customers and more profitable customers, and that just screams for more data and more lending automation. 

Business owners now have to promise the relationship. Most banks have tightened on credit but are still making credit for their existing customer base. So if a bank can have at least some deposits or other income, that can help the customers with extended credit. 

What blind spots do you believe bankers commonly overlook when making lending decisions, without having all the technology in place?

It’s not about having technology in place but about having the right technology in place and the right architecture overall. Most banks have their data stored in a bunch of different silos — a lot of spreadsheets, some databases, and maybe also some databases from a partner vendor. And even if the data is in the core system, it’s hard to access. 

There’s a lack of proper data strategy overall and a lack of the right tactics being driven off their data strategy. For example, banks have a rich source of data in their credit underwriting but normally aren’t making any real quantitative decisions or using a score that can be compared quarter to quarter or month to month.

Another thing is making sure to collect this data in one spot and create a data warehouse, with a single data model and language across the bank. Whatever product you are dealing with, you need to have a standard way to aggregate your data, and then the world is your oyster. It is in fact, the new oil at data. We can make all sorts of lending, profitability, and relationship decisions. We can help the customer so much better if we know what they’re doing and have that 360-degree view. But it all stems from having the right strategy.

How do you see the use of AI within the future of decisioning, and how do you think regulators will deal with it?

AI is huge and is definitely the future in banking. I call it the long tail, as it’s a collection of a lot of little things that may not have the biggest factor in lending, but the collection of the data happens to be really important. It’s the small things — which we don’t have the time, technology, or strategy of tactics to look at — that now become really important.

In terms of the regulators, I think that they’ll want to see first and foremost that you have a documented process and that your model has gone through governance and review. If banks have that in place, I imagine the regulators will be supportive. What we want to make sure is that the model and the use of AI are accurate, transparent, and free from bias. 

As we move away from pure relationship-based lending and introduce more digital interactions, how can banks use data to identify underserved lending opportunities? 

The fact that we’re having this conversation is a step in the right direction. The biases are definitely present both in terms of how we ask the credit question and how we leverage AI to gather the data. As we’ve seen in generative AI and large language models, if you have biased data and ask a biased question, you end up with a biased answer.

The hard part is that you don’t really know what the AI engine is doing and aren’t aware of the biases in your question. The data, the data model, and the question all have biases in them, so you end up getting a biased answer. 

As we’re asking the questions and testing for biases, the models disclose biases better. It’s to the benefit of the underserved market that we get our analysis more granular about who’s truly a credit risk and who’s not, versus who just gets tainted with being a credit risk.

Until we’re able to eliminate hallucination from the equation, what do you think would be the major impact? 

Any model has the potential for inaccuracies and should be treated the same, running the proper testing regressions and ensuring that the data is accurate. Only as we run in parallel and get to see how accurate a model is do we gain more confidence. It improves at a faster rate than humans do, and I personally have a lot more confidence in models than the average person because of that. 

I trust more AI-driven medical decisions, for example, since I know they’re based on a bunch of peer-reviewed studies and not only one opinion. For example, every year I get an AI-driven MRI scan. It’s way more accurate, but as a trade-off, you get a bunch of false positives. The good news is that humans wouldn’t have caught that, so it does a better job of detecting anomalies. The downside is that now you have more false positives to deal with. The same does for credit and any decision driven by AI. It’s part of the game, and it gets better over time.

In every revolution, people have been afraid of losing their jobs. Take for example the Industrial Revolution and how people feared it would increase unemployment rates. Do you sense any similar fears? 

I sense exactly the same fears, and it’s a common argument, but I kind of laugh a little bit about it. I always challenge anyone that brings up this fear and ask them to show me where technology replaced human work. It has never happened in history, but in the short term, it pushed people to retrain. The question that should be asked then is, “What are the new skills to handle AI?”

The future is human expertise combined with AI, as proven in chess and a number of other endeavors. I think we’ll actually employ more humans leveraging AI, and I think those humans will be much more accurate, much more productive, and much more effective at their jobs, particularly in banking and credit.

An MIT study came out a couple of months ago showing that employees are more satisfied with their jobs not only when they don’t have to do manual tasks, but also when knowing they have better answers. Using these AI models to look at 400 or 4,000 pieces of data and not just one gives me more context when I have to sit in front of a customer or partner and talk about decisions. We’ve analyzed much more than we ever have in the past, and it gives me more confidence and a deeper understanding, and makes my job better. 

Do you think smaller banks will be able to adapt to the new status, and what will happen to the ones that don’t have the budget, resources, or knowledge to adapt?

Every bank has resources; it’s just a question of where you direct them. For smaller banks, it’s even easier to have an outsize impact because they can make changes faster. 

You’ve got to start where you stand. Whatever you can do, whether it’s getting educated, trying a new system, or experimenting, and whether you’re a $30 million, $300 million, or $1 billion bank, it can be as effective across the board.

Also now, unlike in the past, you can buy and consume data management tools and AI in particular on a per-use basis, so it’s almost the same price for various institutions, as you’re only paying for bandwidth or processing power.

And yet, smaller community banks are still pen-and-paper driven. What do you think would be their change catalyst? 

It’s a race against time. It’s now table stakes for banks to operate, and those banks that don’t want to operate will get acquired. Banks that have the technology, training, platform, and AI will be in a perfect spot to get acquired because they’re in a better position to break even and handle that faster.

I think it’s the haves and have-nots when it comes to technology, and I’d rather be a $30 million bank just starting off with a huge advantage, allowing the bank to be in a position to dominate in the future. As you pointed out, many banks are not going to get there, and it’s a race against time.

Let’s close with some predictions about the small business lending landscape in the next 12–24 months (just your personal opinion, not binding or presenting any advice, especially not on the investment side).

Don’t take my advice as anything more than just one man’s opinion. But I’m really optimistic about small business lending. Having more technology platforms out there makes it more efficient, and banks are able to lower their cost structure, which means decreased pricing. I also think more banks are focused on the underserved small business segment in general, making it even easier for the small businesses.

Some of the technology allows any bank to connect to ERP systems or accounting systems, and that’s a huge plus. We’ll get better credit information, take less of the small business’s time, and be able to deliver better insights overall, both in terms of standard operating and in terms of credit risk. And that goes on and on. So I think the next five years will be revolutionary for small businesses.

To have this take place in America, which is already a great entrepreneurial country, is fantastic. We need to pick up some of the things that are happening around the world in terms of payments and AI, and build a responsible, ethically-thought-through application to continue to help our customers as banks and as small businesses.

Do you think 1071 regulation will have an impact in the long term, positive or negative, on the small business lending space?

In the short term, it will have a negative impact. It’s just one more thing on our stack that we have to spend resources on. 

But in the long term, the intent is right, and it’s better to gather the data and make decisions accordingly. So as long as it’s done correctly, with accountability, and with the correct governance, I think it will have a positive impact.

Chris Nichols - Bio

Chris Nichols is Director of Capital Markets for SouthState Bank, a $45B publicly traded community bank in the Southeast. In addition to capital market activities, Chris supports loan pricing, an SBA line, payments, innovation and fintech investing for the Bank. Chris produces the popular Banker-to-Banker blog and is a frequent host on The Community Bank Podcast.

Knowledge is power

Unlock your

full lending potential

Transform your business lending

Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.
Become a Lama AI insider
Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.
© Lama AI Inc. 2024
We use cookies to ensure the best possible experience. By using our site you agree with our Cookie Policy