The Utility Trap: What Actually Makes a Bank Convenient?

The afternoon convenience changed (0:00)

On 27 June 1967, Barclays unveiled a cash machine outside its branch in Enfield. Reg Varney, the star of On the Buses, became its first public customer. He inserted a carbon-14 cheque, entered a four-digit PIN and withdrew up to £10. Before that machine, the bank was open when the bank chose. Now, it was open when the customer needed it. Parts of the origin story, including who invented the machine, are disputed. But the machine was real. In one afternoon, convenience meant something new. Then the cash machine stopped being special and became something customers simply expected.

The utility trap, in one number (0:49)

More than 100 million UK current accounts exist, but only about one million switched during 2025. Banking is an essential service. Everyone needs to have one; otherwise, you can't get a job, you can't get credit, etc. You can somewhat opt out, but you would struggle significantly. While some people might see banks more as a commodity, I see them more as a utility. It may not be a utility in how it operates, but a utility in how customers perceive it. The trap is becoming essential but invisible. I am not claiming that price does not matter, or that anyone escapes it forever.

That difference between operating like a utility and being perceived as one matters.

Jamie Dimon is right, in theory (1:40)

In 2016, He said utilities are monopolies with regulated prices and returns. Banks compete, judge risk and sometimes turn customers down. He concluded there is nothing about banking that remotely resembles a utility. But in the same letter, he wanted know-your-customer checks in “a utility that the entire industry can use”. The plumbing could be a utility, but the bank could not. In intellectual theory, banks are not utilities, but in practice, customers may see them that way. He describes the institution. I am interested in the customer's behaviour. And the plumbing becomes obvious when you actually map what moves the money.

What sits underneath your bank account (2:37)

Under a UK bank account sit three core payment systems Bacs, Faster Payments and CHAPS. Bacs carry salaries, pensions and Direct Debits. Faster Payments carries near-instant transfers, including money sent to friends. CHAPS carries same-day, high-value payments, such as completing a house purchase. Swift sends payment instructions between banks, mainly across borders. It's the CHAPS payments, it's Swift payments, it's all of these rails. In 2025, those three UK systems carried more than 12 billion payments. Toast's and Shopify's small-business loans in the US are made by WebBank, a bank most of their customers have never heard of.

The Uber Pro Card is issued by Evolve Bank & Trust. The bank increasingly sits behind the scenes, supporting the brand that the customer sees. Customers only notice the plumbing when it stops working. I thought that convenience was part of that utility experience, but the research changed my view.

The first part of my argument that changed (3:50)

I started this research thinking that convenience was part of the utility experience. Maybe I'm wrong, or maybe convenience is not a utility characteristic. It's the price, the perks—really just like the calculations. Competing on price keeps you in the utility game. Competing on convenience can take you out of it. Convenience is meeting where the customer is in your marketing, branding, distribution channels and interactions. But convenience keeps being redefined. Today's advantage eventually becomes tomorrow's plumbing. The question is where the customer's need begins, and whether the institution remains useful when that moment moves.

My acquisition theory lasted one data check (4:42)

Navy Federal grew from 6.35 million members in 2016 to 15.52 million members in 2026. My first explanation was acquisition. I thought Navy Federal acquired others and became the convenient option because of its scale. The merger records said almost the opposite. At most 0.02% of its growth came through mergers. More than nine million additional members were not transferred in. They chose to join. So acquisition does not explain Navy Federal's growth. The numbers did not weaken the acquisition theory. And they removed it. Navy Federal's growth was chosen, not bought.

If acquisitions did not explain it, perhaps marketing did.

Spending $228.7 million was not the answer (5:31)

Navy Federal spent $228.7 million on advertising and promotion in 2025. I wondered whether utilities spend little on marketing, whilst Navy Federal was spending more and therefore breaking out. US electric utilities spent about 0.07% of revenue on advertising. Credit unions spent 1.91% of income, while Navy Federal spent 1.52%. So I was right that utilities spend very little on marketing. But Navy Federal was below the credit union average. Among credit unions over $10 billion, marketing spend told you almost nothing about member growth from 2020 to 2025. The correlation was 0.06, and no size band rose above 0.2.

So maybe I'm wrong then? Navy Federal did not win simply by buying more attention. The amount spent was impressive. The rate of spending was not unusual.

That left the most visible difference.

The branch at the first pay packet (6:44)

Branches. At two military recruitment depots, Navy Federal managed to recruit almost every single graduate into a member in 2025. Navy Federal has the only credit union branches at those two recruit depots. Recruits need an account for electronic pay. Navy Federal is there when that need begins. Maybe it was because of the branches. But other credit unions have branches, and they don't seem to be succeeding in the same way. Among credit unions with 50 or more branches, starting branch count had almost no relationship with member growth.

The branch alone was not the answer. The timing and purpose of the branch mattered.

Then I wondered whether the real advantage was having a monopoly in a dense location.

A branch monopoly does not guarantee growth (7:39)

Federal rules allow no more than one bank and one credit union on a US military installation. In the data, 139 of 142 identified base ZIPs had exactly one credit union on site. Almost every base credit union gets that local monopoly. Yet holding one predicted almost nothing about growth.

Fort Sill's credit union had a branch at an Army boot camp and still shrank 15.7%. Navy Federal held 42 of those 142 identified installation ZIPs. PenFed held nine. Maybe Federal is unusual because it appears at many entry points. But a captive location is not enough. The institution has to remain useful after the member leaves. That second half changed my conclusion about what Navy Federal was doing.

Convenience follows the customer (8:39)

PenFed had no boot-camp branch and still grew 91% over the decade. So a boot-camp branch is not the whole method.

Navy Federal is meeting its customers where they are, and it works for them in person. But it might not work for everyone. It depends on who your customer is. Navy Federal can see pay dates, training and which graduates are upcoming. During a 43-day government shutdown in 2025, it paid $352.4 million in advances to affected families. Convenience is being there when the need starts, being useful when the needs move, and then anticipating that next need.

Monzo wrote its growth plan in one paragraph (9:31)

In its 2018 annual report, Monzo said 80% of its growth came from word-of-mouth referrals. The average customer already had 15 friends on Monzo. In the same document, Monzo said it would add features that worked better when more friends used it. That's more specific than asking customers to recommend an app. It's probably a piece around product development leading to word of mouth. The feature would create a reason for another person to touch the product. Product development became distribution. The customer carried the invitation.

The clearest example happens after dinner, not inside a bank.

A walkie-talkie needs two people (10:19)

In its 2025 financial year, Monzo customers made 8.8 million bill splits and 10.1 million payment requests. One walkie-talkie on its own is useless. You hand the other one to a friend. Monzo built bill splitting, shared tabs and easy payments into the product. A bill split only works when another person receives the request. Why did that create word of mouth? That's actually a convenience play, like bill splitting and sending money. Monzo met the customer when money became social. The feature did not merely serve the customer. It helped distribute the product.

But the strongest case against this argument is that price still moved the money.

Price did not disappear (11:09)

In 2024, Monzo's instant-access savings pots took £5.4 billion, more than the entire £5.2 billion rise in deposits. That's it. Monzo's word-of-mouth growth does not prove price stopped mattering. Its fee-free spending abroad offered a price advantage. The nuance is that Monzo, Revolut and others don't always compete on price. Customers may arrive for convenience and move more money when the rate improves. Monzo also never published how many customers joined through bill splitting. Paid referrals later represented around 14% to 17% of growth. Price is not irrelevant. It is one part of the playbook, not the whole explanation.

That leaves a useful link between a military credit union and a neobank.

Macro moments and micro moments (12:03)

Navy Federal captured 92% of Great Lakes graduates in 2025 during a major life moment. Banks are thinking very macro, like when you first get a property or first start a job. Monzo meets smaller moments, a restaurant bill, a shared holiday tab or the days before payday. Product features don't need to be digital, though. Navy Federal's feature was the monopoly of branches when someone first got a job. Monzo's feature was bill splitting when someone shared a meal. A macro moment might bring a customer in once. A useful micro moment can bring the product back into their hand repeatedly.

Five out of twelve (12:47)

I studied a credit union with tens of thousands of members. I tested what a member could actually do against 12 everyday convenience checks. It scored five. Monzo scored 12. The credit union has no card, so there is no Apple Pay or in-app card freeze. New members completed a 30-field web form and wait up to one working day. Its app uses a shared white-label platform. The institution I studied is trapped inside a provider which controls much of the customer experience. It owns no big life moment and no small everyday one.

It mostly waits for customers to arrive.

The obvious villain is the technology provider, but the evidence is a lot less comfortable.

Technology is not the whole villain (13:38)

A small credit union I studied lacks in-app joining. The obvious villain is its app provider. But that provider offers in-app joining and an eligibility check. The credit union had not switched those features on. So the provider was not the whole problem. Building everything yourself is not a simple answer either. Another small credit union developed its own lending platform, said it strained its resources, and fell below its capital minimum. Small institutions can be quicker because they have fewer people. But speed without a clear priority can damage the institution.

Technology cannot decide which customer moment deserves the investment.

The Convenience Checklist (14:18)

We built a free convenience checklist with 17 tests. It is for banks, building societies and credit unions. The first section checks macro moments, such as joining when a new stage of life begins. The second checks micro moments, including splitting a bill or getting help before payday. The third checks table stakes, including Sunday payments and freezing a card in the app. They might all be features for a moment, but they're all table stakes now. Download it free from the Doshi Score what customers can actually do, not how attractive the app looks.

Even a perfect score only measures the definition of convenience we have today.

The undertaker who removed the operator (15:05)

In Kansas City, the story goes an undertaker called Almon Strowger believed telephone Operators were sending customers to a rival. The details are disputed, but the problem he saw was simple. The popular story says his rival's wife was an operator who sent callers to her husband. The earliest accounts only say Strowger blamed the operators. In 1891, Almon Strowger received a patent for an automatic telephone exchange. His underlying problem was clear. Whoever controlled the connection could influence which business got the customer. Strowger replaced that connection with a machine.

Owning the infrastructure did not guarantee owning the relationship. The layer in front could still decide where the customer went. For a bank, whoever sits between the customer and the business can steer the customer.

Today, that layer could be ChatGBT, Claude or Meta's Muse or another AI agent.

The new operator is already arriving (16:09)

Since May 2026, some US ChatGPT users have been able to connect financial accounts through Plaid. It can read spending, subscriptions, investments and upcoming bills. cannot yet move money. In July 2026, Visa said AI agents had completed live purchases for customers of more than 30 European issuers. Tomorrow, your customer's AI agent may become the new switchboard operator; the thing that sits between them and their bank and decides where they go. It could be ChatGPT, Claude or whoever they use at the time. The agent is no longer only a prediction.

It is starting to sit between the customer, the bank and the merchant. Banks can only make sure it can reach them.

I can already see an early version of that behaviour in my own accounts.

I am already choosing for the agent (17:06)

But it still cannot move the money.

I am starting to use banks that have really good developer tools, like Monzo, Starling and Trading 212. I'm plugging that into my agent and safely being able to inspect and understand my finances. I use it for personal financial management, planning, goal setting and scenario planning. It can figure out the intent and map different scenarios. I might be a small type of customer, but potentially that could grow even bigger.

The ordinary version is someone asking their phone to organise their money. If a customer sends a machine instead of opening an app, convenience changes again.

The next definition of convenience (17:57)

More than 30 European card issuers had already supported live AI-agent purchases by July 2026. Tomorrow, convenience might mean having a really good API surface and having unique data. Data only that bank has, worth an agent reaching for. People could use their agents on top of your data. If agents choose banks, convenience becomes documentation and thinking deeply about how agents choose. The customer may never open the bank's app. The bank remains essential, but the agent becomes the visible face. The next convenient bank may not have the customer's favourite app.

It may be the bank their agent can understand and reach.

There is no permanent escape (18:42)

A bank can be essential and still become invisible to the customer. Convenience keeps moving from a place, to another person, to a machine. Today, it might mean having a mobile app, but that is far behind what convenience now means. Maybe the whole thing is anticipating what will be convenient, not only what is convenient now. It is all about size plus culture adapting to what convenience is. The difficult part is not adding the last convenient feature. It is anticipating what customers will find convenient next. In 1967, a cash machine let customers withdraw £10 when the branch was closed.

By 2026, AI agents were reading bank accounts and making live. There is no permanent escape from the utility trap. There is only staying ahead.