Stake sizes on UK-licensed online slots fell by an average of 14.2% after 9pm during the first eight weeks of a trial that replaced free-text stake entry with fixed, rounded bet buttons, according to data from six operators covering roughly 1.9 million sessions. The drop was concentrated almost entirely in the £1–£5 band, and it did not disappear when the button layout was changed — it held at 13.1% across the second half of the trial period. The finding lands awkwardly for an industry that has spent two years arguing stake limits would simply push players to unregulated sites rather than change behaviour.

What the trial actually changed

The intervention was deliberately narrow. Instead of a text field where a player could type any stake from 1p to £50, the trial operators served a row of buttons: 20p, 50p, £1, £2, £5, with a "more" option that opened the text field anyway. Nothing else changed — same games, same RTPs, same bonus terms, same session limits. The 20p floor was the same as before.

That last detail matters. This was not a stake cap. Every player who wanted to stake £10 a spin could still do so with two taps. The 14.2% reduction therefore isn't a compliance artefact of a hard ceiling; it's a friction effect. Removing the ability to type £1.37 or £2.80 nudged players toward the nearest round number below their previous average, and enough of them stayed there to move the aggregate.

The 9pm threshold is the more interesting variable. Pre-9pm sessions showed a 3.4% reduction — within noise for a sample this size. Post-9pm sessions showed 14.2%. The gap suggests the effect isn't about the interface at all in the early evening, when most players are staking small amounts anyway, but about what happens when stakes typically rise.

The late-evening stake curve

UK operator data has consistently shown a stake escalation after 21:00. Average spin values roughly double between 19:00 and 23:00, and the proportion of players using auto-spin for more than 100 spins rises from around 11% to 29%. The rounded-button trial appears to have flattened part of that curve, not by preventing escalation but by removing the increments players use to escalate gradually. Moving from £1 to £2 is a button. Moving from £1 to £1.50 no longer exists as an option.

Why rounding works, and why that's uncomfortable

Behavioural research on payment interfaces has known for years that removing arbitrary increments reduces spend. The mechanism is usually described as a combination of anchoring and reduced cognitive load: the button row presents five options, and the middle of five options becomes the default. In the trial, £1 was the third of five buttons and absorbed the largest share of migrated stakes — 41% of players who previously staked between £1.20 and £2.50 ended up on £1.

That's a real behavioural change, not a reporting one. The operators involved confirmed that gross gaming yield per session fell in line with the stake reduction, and that session length did not increase to compensate. Players did not spin more to make up the difference. They spun the same number of times for less money.

Which is precisely the problem for anyone who wants to frame this as a neutral UX improvement. If button design can cut stakes by 14% at the time of day when harm is most concentrated, then button design is a harm-reduction tool. And if it's a harm-reduction tool, the question of who decides the button layout stops being a product question and becomes a regulatory one.

The counter-argument, fairly stated

The operators running the trial have been careful to note three limitations. First, the sample skews toward casual players; high-stake players were underrepresented at 6% of sessions, and their behaviour may differ. Second, eight weeks is short — novelty effects can decay. Third, and most importantly, the trial cannot measure displacement. A player who finds £2 buttons annoying at 10pm on a regulated site may simply open an unlicensed site in another tab, and that player is invisible to this dataset.

That third point is the one the industry will lead with, and it deserves a straight answer rather than a dismissal. The honest position is that we don't know the displacement rate, because nobody has measured it properly. What we do know is that the UK Gambling Commission's own figures show unlicensed site usage has not risen materially since the £2 slot stake limit for under-25s came in during 2024, which weakens the displacement argument but doesn't kill it.

What this means for the stake limit debate

The £5 maximum stake for online slots, in force since 2025, was the headline intervention. This trial suggests the headline intervention may be the less effective one. A £5 cap binds only the small minority staking above £5. A button layout binds everyone, at every stake level, and appears to work hardest exactly when stakes are highest.

That reframes the policy conversation. Instead of arguing about where the cap sits — £2, £5, £10 — regulators could be arguing about interface defaults, which are cheaper to implement, don't require cross-border enforcement, and don't create a two-tier market between licensed and unlicensed operators in the same way a hard cap does.

There are obvious objections. Prescribing button layouts is more intrusive than setting a number. It's harder to audit. Operators could game it by making the "more" option more prominent, or by reordering buttons so the highest stake sits in the middle. The trial itself showed that button order matters: when £2 was moved to the centre position in week five, average stakes rose 4.7% within three days. That's a small number, but it's a lever, and levers get pulled.

Where the numbers point

The 14.2% figure will be quoted for years, in both directions. Expect it to appear in consultation responses as proof that soft interventions work, and in operator submissions as proof that soft interventions are unpredictable and shouldn't be mandated. Both readings are defensible from the same dataset, which is usually a sign that the dataset is measuring something real.

The more useful question is what happens at 9pm specifically. If the late-evening spike in stake size is driven by a small group of players in a particular state — tired, chasing, or simply less inhibited — then an interface that interrupts the escalation is doing something a stake cap can't. It's not stopping anyone from staking £5. It's making £5 require a decision rather than a reflex.

Whether that's a legitimate regulatory goal, or an unacceptable amount of social engineering to build into a spin button, is not a question the data can answer. But it's the question the next round of consultations will have to.