Two words sort British investors by the decade they arrived in rather than by anything on their statements. Put the bills beside each other and the categories come apart.
Which of the two words did you learn first?
Ask somebody where they hold their investments and the answer dates them. Start before smartphones and you say broker, meaning a firm with a phone number and a paper statement. Start after and you say app. Nobody chose that vocabulary: it settled on whatever was newest when each cohort arrived, which is how a word for a screen ended up doing the work of a word for a business.
Neither word describes a way of charging money. A broker is a firm holding permissions. An app is a channel for reaching customers. They feel like alternatives because they arrived in different decades and sold to different people, and Britain’s investing boom coincided with the handset, so the interface got mistaken for the product.
The labels sort people, and people are not a price list
One number in the FCA’s Financial Lives 2024 survey settles this faster than any argument about fees. Among UK adults, 5% of men were using a trading app against 1% of women. Five to one, on a product both are quoted identically and both can open in the same afternoon.
Note what the figure does not carry: it counts app users, not investors, so it says nothing about who holds what elsewhere, and no published breakdown settles whether the gap narrows further up the wealth distribution.
No fee schedule explains either figure. Both sexes are quoted the same commissions and offered the same instruments, and a five-to-one split survives that, so what is being sorted is who was persuaded rather than who is charged what. When somebody says apps are cheaper, they usually mean a marketing posture adopted by firms courting a particular cohort. Postures are revisable. Terms and conditions are what bills you, and they sort differently.
The cheap part is the part you were shown
The concession first, because it is real. On the number both sides advertise, dealing commission on a UK share trade, the app-first firms forced the price down and the established names followed. Minimum deposits collapsed with it. Trade British shares a few times a year and hold nothing else, and the app-first arrangement is genuinely cheaper.
Now notice what that convergence did to the comparison. A charge printed in every table gets competed towards zero, and what remains sits where nobody tabulates it. The category was built to describe a pricing gap since competed out of existence, so the label has outlived the thing it described.
Buy an American share from a sterling account and something turns pounds into dollars, twice, once going in and once coming back. It is charged as a percentage rather than per trade, and nothing leaves your balance in a way you can point at. The rate you receive is worse than the one on the screen, by a margin stated in the terms and almost never on the ticket.
Which makes conversion the one charge you cannot check at the moment you incur it. Every other line can be read off a confirmation. This one is established afterwards, by looking up a mid-market rate the platform had no reason to show you and doing the subtraction. Nor does it sort by label: conversion pricing follows a firm’s own funding arrangements, invisible from outside and revised without notice.
Neither label tells you who charges for stopping
The received version says app-first firms recover their margin from inaction, through monthly charges once an account goes quiet and flat fees for taking money out. The funded accounts declined to cooperate. Across the FCA-regulated UK platforms opened for testing, the monthly charge for going quiet turned up at two of the oldest names in British retail trading, and the flat charge for taking money out sat with the app-first one. The direction is the finding; the amounts are another article.
There is a reason for the split, and it has nothing to do with interfaces. A charge on inaction pays only where a firm has a long book of half-forgotten accounts to levy it on, which takes decades to accumulate. A charge on the exit pays sooner, on a book that turns over fast. The two charges encode the age of the customer list rather than the age of the software, and any firm that lives long enough acquires the first kind.
| Cost line |
Quoted before you open? |
When you meet it |
Does the label predict it? |
| Dealing commission |
Both, prominently |
First trade |
Partly, less every year |
| Bid-offer spread |
As a from figure, if at all |
First trade |
No |
| Currency conversion |
In the terms, as a % |
First non-sterling trade |
No |
| Overnight financing |
In the terms, as a formula |
The second day of a position |
No, it follows the product |
| Dormancy charge |
On the fee page |
Months after you stop |
No |
| Withdrawal charge |
On the fee page |
The day you leave |
No |
How UK platforms present these charges, August 2026.
The fourth column deserves less confidence than a column tends to attract. Nobody publishes a category-level average of these charges, so a no there means only that the labels failed to predict the direction on a handful of funded accounts: fair warning against the shorthand, nothing like proof it is always wrong.
How would you price this honestly?
Not from a pricing page. Every charge that still separates these firms shows up on a statement rather than in a fee table, so the only way to compare them is to become a customer of both, deposit, trade, wait and withdraw. That is the method behind the comparison of UK trading apps, which is assembled by opening and funding live accounts with its own money rather than compiling rankings from providers’ published fee schedules, and its trading app comparison is where that work lands.
That method has a limit which this article walks straight into. Funding an account produces exact figures for the firm that was funded, and the comparison attempted here is between two populations. A handful of bills cannot support a sentence beginning with the words apps typically. They can only show that firms sorted by interface fail to line up by cost, which is the honest ceiling of a method that samples one account at a time.
The register has no field for an interface
Britain keeps a public register of every firm permitted to do this, and it is worth noticing what it records: permissions, status, addresses, trading names, the individuals accountable. There is no field for whether the firm has an app. An interface is not a regulated attribute, because a screen is a delivery mechanism sitting on permissions granted for something else entirely.
Which is why so many firms are now both. Long-established brokers shipped apps years ago, and app-first firms acquired the permissions to offer the full product set, so sorting the market into two boxes means deciding where a firm occupying both belongs, and the answer usually turns on its founding date. The register records trading names against firms, too, so one authorised entity can carry several consumer brands and a reader comparing two of them may be comparing two front doors on one permission.
There is a pattern here worth naming, because it will repeat. Every cohort’s new interface becomes the next cohort’s incumbent. The firms that arrived app-first are no longer new: several are listed, hold full permissions and run dormancy policies of their own. The word app is ageing into the word broker, and in a decade something else will arrive to be called the opposite of both.
So there is little point choosing between two labels that keep swapping contents. The property worth judging a firm on is the one neither label has ever encoded: how many of its charges you can find, in money, before handing over any. Conversion, financing, dormancy and the exit are the four that separate firms, and one publishing all four plainly has told you more than any argument about interfaces will.
…
