Apple Card Fail?

Years and years ago, I did a business process analysis for an insurance company. One of the people I met there and interviewed said:

“We’ve bred a group of box tickers and button pushers”

This was, in context, about their underwriters. They ticked boxes and pushed buttons. Their process was black and white: you’re either accepted or rejected. It was clear that this was not right as you can replace box tickers and button pushers with software. Equally, and more importantly, it was not right for the customer, as often there are gray areas.

Customers who fall between the lines.

That’s the issue and this strain of thought came back to me as I read this:

[With Apple Card] Goldman implemented an algorithm that either accepts or rejects applicants within seconds. If your application is successful, you’re offered personalized credit terms such as those shown above. If not, you receive an instant rejection letter without an equally fast route to appeal.

Hmmm. Not very Apple.

When you automate or implement a process which delineates between a pure yes and a pure no, you have issues. Sure, it’s efficient but it’s not appropriate. It leaves many between the lines.

I’ve been through this myself. I’ve been in a position of having a new job, moving home, changing life and being rejected because I’ve not been in the address for a year, not having a job for a year, too much change.

Most financial firms are incredibly simplistic: they just want to check the buttons and tick the boxes. And yet, in today’s world of emerging artificial intelligence, you can and should be far more nuanced.

I mean, if you take the example of someone who’s just moved, changed address and changed job – a situation that is highly likely when someone switches companies – why wouldn’t you give them credit? The algorithm says: new address, different employer, no credit. But then the algorithm is wrong.

In a similar context, I have so many friends who have travelled and changed their situation, but the algorithm, the box ticker, the button pusher does not accept their new situation. Computer says no.

The real issue here is identity, as David Birch would say. Why do we have so many issues with onboarding customers, giving them credit, providing them with trust.

That’s the core issue: trust.

I often talk about banks acting as trusted stores of value, but think about it the other way around: banks trusting you with their money. Giving you credit. Can we trust you?

Banks and their customers are in a trust partnership: I trust the bank to look after my money and the bank trusts me to pay it back. This is the basics of banking, insurance and financial services. It’s all about risk management.  The risk of them losing my money and the risk of you not paying the money back.

And yes, you can put this risk metric into an algorithm, but it needs space and latitude. Space and latitude for where it’s not clear, where it’s not black and white, where it’s not a straight A and B.

This is where the industry has always had a challenge. For efficiency, the financial markets prefer a straight yes or no. Where’s the maybe?

Related: Should We Gamify Banking?