The only part of a British bank's service that improved this year is the overdraft.
That is not a joke about priorities. It is what the August 2026 figures show, and the figures are not ours. Twice a year the Competition and Markets Authority requires the seventeen largest current account providers in Great Britain to publish independent service quality scores, measured by Ipsos across roughly a thousand customers each. Four dimensions, asked separately of the same people: the provider overall, its online and mobile banking, its overdraft, its branches.
Between February and August, the market median rose on exactly one of them.
The number that should not have moved
Overdraft is the only dimension where the market median rose between February and August, from 63.5 to 64.5. Overall satisfaction was flat, and both branch and mobile scores fell.
Ten of the sixteen providers publishing an overdraft score improved. The largest single move in the entire dataset belongs to NatWest, up nine points from 56 to 65, and it is not isolated: Royal Bank of Scotland, the other brand in that group, rose four, while all three Lloyds Banking Group brands stayed where they were. Two brands sharing infrastructure and pricing policy moving together looks like a group-level decision rather than a brand campaign.
Overdraft is the least fashionable product in retail banking, and the one the regulator reshaped most recently. The FCA's 2020 pricing rules banned fixed daily and monthly fees, stopped firms charging more for unarranged borrowing than arranged, and forced every provider onto a single annual rate. It moved. The channel that receives the investment and the attention did not.
The channel is converging, and the table is missing the comparison set
Mobile banking scores are converging. The spread across the seventeen fell from 30 points to 25 in six months. The bottom of the table rose from 57 to 60, the top slipped from 87 to 85. Over the same period branch scores diverged, moving from a 29 point spread to 32. This is not general levelling, it is specific to the digital channel.
The movement is not evenly spread. Chase lost four points on overall satisfaction and dropped out of the top three for mobile. Monzo lost three, Starling one. The Co-operative Bank, last on mobile in February, gained three.
One explanation covers both ends. The challengers scored where they did partly because what they offered was unfamiliar, and unfamiliarity is not a renewable resource. Six years of the rest of the market catching up moves what a customer treats as ordinary, and a score measured against expectation falls when expectation rises, even if nothing about the product got worse. At the other end, the laggards are climbing a much lower hill.
The warning: two waves is not a trend, and the next data point does not arrive until February 2027.
The more interesting limitation is who is in the table at all. Participation is defined by holding a licence to offer personal current accounts. Revolut operated in the UK as an e-money institution until 11 March 2026 and therefore sat outside the regime entirely. It has thirteen million UK customers, most of them on e-money accounts rather than current accounts while this wave was in the field.
So the convergence is real, and it is convergence inside a closed group. Your customer is not comparing your app to Bank of Scotland. Seventeen banks are converging on a standard set by a group that no longer contains the products their customers use to judge them. Call that a playing field levelling out if you like. It looks more like a group agreeing on a benchmark the market has already left.
If the channel is not the differentiator, what is
Having a mobile channel stopped separating banks some years ago. Every one of them has an app, some better than others, and the same is now true across most of the economy. Whatever distinguishes two banks today, it is not the existence of the thing.
The answer most of the market is currently reaching for is AI. Agents, in-house models, a roadmap slide carrying both. I am sceptical, and the reason is structural rather than technical. Everyone started at the same time. A capability the whole market acquires on the same timeline commoditises within a few years, and then it is simply the cost of staying in the room.
The differentiation I would back is narrower and much less fashionable. Take one segment, find what it cannot currently do, and rebuild the relevant part of the product around that. An SME bank could pick Shopify sellers, look at what breaks in their week, and build for exactly that, integration included. It is a smaller claim than being an AI-first bank, and it holds for longer, because it rests on a problem rather than on a technology.
Revolut is the cleanest illustration. It launched in 2015 as a prepaid multi-currency card at interbank rates, for people who spent abroad often enough to notice they were paying a three to five percent markup every time. One segment, one specific cost, one product built to remove it. Everything else came afterwards.
Where separation actually comes from
The seventeen will keep converging, because they are solving the same problems with the same tools on roughly the same timeline. Nobody in that group is going to pull ahead by shipping a better version of what everyone else is also shipping.
Separation comes from choosing a group of customers whose problems nobody has bothered to solve properly, and then solving those. It is a smaller ambition than most roadmaps carry. It is also the one that survives a competitor reading your announcement, because a competitor can copy a feature far more easily than they can copy a decision about who you are for.
Before we build anything, we run a scoped review to work out what a client is actually competing on. Sometimes it ends with us saying the build is not the answer. We treat that as a result.
