The billion-dollar business case for UX in banking
User-centred design reduces costly banking errors by designing systems around real human behaviour and limitations.
UX and human-factor design could provide a return on investment of at least $1bn to a global bank over a five-year period. Not only that, it could provide an almost incalculable reputational boost.
Sounds like quite the claim, doesn’t it? The promised returns make saying “sign me up” very straightforward.
There is just a small twist – the return on investment isn’t based on additional sales, increased efficiency or improved customer lifetime value. Rather it is based on the somewhat less digestible metric of ‘revenue not lost or squandered’.
I don’t deny that it’s a harder sell, but a dollar is a dollar, whether it arrived in the form of money earned, or money not wasted. The net result is the same.
It’s at this point that I feel compelled to introduce my insurance analogy.
Back when I was running my agency, I would get an annual phone call from my insurance broker dude. He was good fun (in fact, there’s an off chance he might even be reading this) and as well as him trying to sell me as much insurance as he could, we had a good laugh, usually about rugby, or his ongoing attempts to play over-35s cricket.
As I got to know him, I used to tease him every year that his job was to phone me up, scare the s—t out of me about every natural disaster that was inches away from the business, and having established the real and present danger, sell me the insurance that would make it all go away.
“Have you thought about cyber security, do you know cyber incidents are up 70% year on year since 2015?”
“Have you considered what would happen if you were knocked down by a bus tomorrow, could the business go on without you?”
You get the picture. Of course, more often than not I bought the insurance. I had plenty of other things to be thinking about, meaning that the purpose of the insurance was simply for me to have slightly fewer things to worry about as I led the business.
So what has this to do with global banks and a $1bn return on investment?
Well just as insurance can protect a business against losses due to cyber-attacks, or losses due to the unavailability of key leaders, so UX can protect a business against losses due to human error, or as euphemistically expressed in the media, “fat fingered bankers.”
Human factors design principles assert that a system should be designed around the limitations of humans in the same was as it is designed around technical and other design parameters. Humans get tired and distracted and impatient and while it might be controversial to say it, bankers are humans too.
The poster child bank for distended digits has been Citigroup Inc. It seems their fat fingered financial fat-cats can’t give the money away fast enough. Here’s some of what they’ve been up to over the past few years.
In 2020 they paid $900m ($900,000,000) to Revlon creditors (instead of a $1.5m loan interest payment). They asked hedge fund Brigade Capital to return $176m, who unsurprisingly were in no hurry to oblige. The various parties took the whole matter to court, as a result of which Citi got $400m back in 2021, leaving a $500m hole in cashflow.
In May of 2022 a London-based trader erroneously placed an order of $1.4bn ($1,400,000,000) instead of $58m, triggering a crash across European stock markets which immediately cost the bank at least $50m. The mistake triggered a sell-off which wiped out as much as $300bn from the stock markets at its lowest point. As well as the losses, the bank was fined $78m for its troubles.
And just last week they credited a client account with $81tn ($81,000,000,000,000) (instead of $280) before the error was spotted. This is enough to buy the entire US Stock Market and the UK economy. This time, the money was retrieved within a few hours, doubtless to loud sighs of relief inside Citi HQ.
These are only the mistakes which made it into the public sphere. Last year The Financial Times reported that Citi experienced 10 near misses of more than $1bn in a single year.
It is the easiest thing in the world to analyse and judge these mishaps from the safe distance of hundreds of miles, and from the quiet serenity of a study. These mistakes were doubtless made by stressed bankers trying to juggle professional and personal priorities in the midst of a busy day. But it’s precisely because of that stress and busyness that human factors engineering has emerged as critical component of high-stakes software design. By high-stakes I am referring to the technology which helps humans land a 747 in thick fog, diagnose a patient in ICU, or transact billions of dollars.
UX can protect a business against losses due to human error, or as euphemistically expressed in the media, “fat fingered bankers.”
Human factors design principles assert that a system should be designed around the limitations of humans in the same was as it is designed around technical and other design parameters. Humans get tired and distracted and impatient and while it might be controversial to say it, bankers are humans too.
What might that mean for Citi’s travails? While I don’t have access to all of the facts in the various examples, there is enough reported in the press to at least diagnose some problems and offer some solutions.
Solving problems and completing tasks
In the Revlon example, the use case of paying ‘interest only’ to multiple vendors hadn’t been considered and therefore hadn’t been designed. A number of media commentators at the time used the phrase “trick the software” when outlining how bank employees needed to get the job done. This is straightforwardly remedied – design for tasks, identify this as a relevant task, prioritise it accordingly within the system interface, and map task flow.
Understandability and mental models
The Revlon interface is also a horror show. Even the application of established form design principles would improve it beyond recognition. Labels, categorisation, clustering, layout and calls to action could all be significantly improved. Usability is important in all interfaces, but usability is impossible without understandability. Users need to feel confident that what they think they are doing and what they are actually doing is one and the same. Mental model alignment can be easily tested during design and build – using techniques such as perception tests and click tests.
Creating theatre and dramatic pauses
It is very helpful for users to get a summary of what they are about to do before they do it. It is particularly helpful if the user is about to do something important, irreversible or which will cost them money. The most relatable example of this is in the world of e-commerce, specifically the screen you see after you fill your basket, enter your personal, delivery and payment information. This summary screen clearly presents what is about to happen. “You have bought these goods at this price, arriving on this date to this address, costing this.” This page usually has links to terms and conditions and large button saying, “Buy now”. The page has created some theatre and made it crystal clear to the user exactly what will happen when the button is pressed.
Similarly, on email broadcast platforms, if you try to send a campaign you will get a page summarising how many recipients it will go to, and what the campaign is called. This allows the user a final chance to make sure they’ve got everything lined up correctly.
Citi could create a summary page which goes far beyond the “You are about to send €900m, are you sure?” by using graphics, colours and fonts of different sizes to really emphasise when very significant amounts are going to be sent. The system communicating to the user “what you are about to do is a very big deal” is the best way for the user to understand the implications of their actions.
Identifying weaknesses in the system
According to the media, each of the three incidents involved multiple sign-offs, meaning that more than one human did the wrong thing. It’s easy to write this off as “lazy inept bankers are all so useless” but imagine for a moment that you are one of the additional sign-off people who received the request to approve the transaction. Here is what your world is likely to look like as the request hits your desk. It’s 4pm. You need to finish sharp at 5pm to pick one of the kids up from football practice. You’ve just received a text from your partner to collect some groceries on the way home from work. You have promised your boss a report first thing in the morning, and you want to get it wrapped up before leaving. There is an internal political brouhaha developing in your email inbox and you want to stay on top of it. And the approval request arrives in from your colleague Dave. You’ve worked with Dave for a decade, and you know he’s rock solid, rarely makes a mistake, and is a “measure twice cut once” kind of a guy. Dave has sent you a separate email saying he really wants to get the transaction approved by close of business to allow him to help his client who has been applying pressure on him.
How much attention are you going to give that transaction? My guess it’s getting a cursory glance and a pretty quick approval.
This context is the reason that having to change a password every month makes a system less secure not more secure. The busy stressed human getting the password refresh messages makes their password increasingly predictable and guessable over time.
The system needs to be designed to specifically take the approver step by step through the mechanics of what they are authorising, drawing attention to exactly what they are signing off and precisely what happens next if they do.
Improving outcomes therefore is about improving the overall system, not just the technology element. By looking at context, tasks, workflow, humans and technology, the system designer can identify actual strengths and weaknesses across the system and design accordingly.
I’d love to offer my services to Citi to help them overcome some of these challenges. With the voice of my old mate from insurance ringing in my ears, I could limit the budget to as little as €999,000,000, leaving Citi with a cool $1,000,000 return on investment.
