Auto-Play Stops at 25 Spins, but 61% Set It to 100
Roughly 61% of players who touch auto-play in online slots set their spin count to 100, even though most operators cap the feature at 25 spins per session and many licensees have moved to 10. That gap between what the tool allows and what players reach for is one of the more durable mismatches in the U.S. market, and it says less about player appetite for risk than about how badly the default settings on auto-play are designed.
The 61% figure comes from a pooled look at session logs across a mid-size operator's slots catalog in 2023, covering roughly 4.2 million auto-play activations. It is not a published study and it should not be treated as one. But it lines up with what product managers at several U.S. facing platforms describe privately: the spin-count selector is the single most-touched control on the auto-play panel, and the modal choice clusters hard around 100. The 25-spin cap, meanwhile, is a compliance artifact more than a product decision. It exists because regulators and responsible-gambling teams wanted a speed bump between a player and an unattended session, and 25 was a number everyone could agree on.
The 25-spin cap is a compliance number, not a player number
Auto-play has had a strange regulatory life in the United States. In the regulated online market, it is permitted but constrained. New Jersey's Division of Gaming Enforcement, which has been the most detailed of the state regulators on slot feature design, has pushed operators toward session limits and loss limits on auto-play rather than outright bans. Pennsylvania and Michigan have taken similar positions with different wording. The practical result is a patchwork: a player in one state may see a 25-spin maximum with a mandatory stop on any single win above a threshold, while a player in another sees 10 spins with a loss-limit prompt every time.
The 25 number itself has no particular statistical basis. It is short enough that a player is likely to look at the screen again within a minute or two on a typical 1.5-second spin cycle, and long enough that it does not feel punitive. That is the whole logic. It is a human-attention heuristic dressed up as a rule.
What the compliance framing missed is that players do not experience auto-play as a risk control. They experience it as a convenience feature, and they set it the way people set a microwave timer: to the round number that covers the task. If you want to spin while you make coffee, 25 spins is not enough. If you want to spin while you read a work email, 100 is roughly the right length. The 61% cluster at 100 is not a risk preference. It is a duration preference.
Why the selector matters more than the cap
There is a subtle product problem here. When an operator caps auto-play at 25, players do not stop wanting 100 spins. They run four auto-play sessions back to back. The session log shows four activations of 25 rather than one of 100, which makes the cap look effective in aggregate reporting while changing almost nothing about how long the player actually spins unattended. Several responsible-gambling dashboards count auto-play activations as a engagement metric without adjusting for this, which flatters the numbers.
The alternative some operators have tried, a hard 10-spin cap with a mandatory re-confirmation, does interrupt the behavior more reliably. It also produces a measurable drop in slots revenue per session, which is why it tends to get rolled back quietly after a quarter or two. That is the honest trade-off, and it rarely gets stated that plainly in the press releases about "enhanced player protection features."
What the 100-spin preference actually correlates with
The session data behind the 61% figure shows a few patterns worth flagging, though again, single-operator data is not a market study.
Players who select 100 spins are more likely to have a deposit in the prior 24 hours, more likely to be playing a high-volatility title, and more likely to have used auto-play on the same game before. The last one is the strongest predictor. Auto-play at 100 is a habit, not a one-off decision, and habits form fast. A player who runs 100 spins three times on a Friday is likely to do it again the following Friday.
The volatility correlation is the one that deserves more attention. High-volatility slots are exactly the games where 100 unattended spins can produce a swing large enough to matter to a player's bankroll, and exactly the games where auto-play is most attractive because the base game is tedious between features. The feature is most convenient precisely where it is least safe. That is not a coincidence; it is a design tension that no spin cap resolves.
The loss-limit interaction nobody tests
Most U.S. operators now require a loss limit to be set before auto-play can be activated, or prompt for one at the start of a session. In practice, the prompt is where the interesting behavior happens. Players who are shown a loss-limit field before their first auto-play session set a limit at a meaningfully higher rate than players who are shown it only after a loss. That is the standard "commitment device" effect, and it is one of the few interventions in this space with decent evidence behind it.
The problem is that the loss limit and the spin count are set on the same screen, and the spin count is the field players actually care about. When a player types 100 into the spin field, the loss-limit field tends to get whatever number makes the panel go away. The order of the fields matters. Put the loss limit first, make the spin count the secondary decision, and the completion rates change. Very few operators A/B test this, because the compliance team owns the panel and the growth team owns everything else, and nobody owns the interaction between them.
The U.S. market is moving in two directions at once
The regulated U.S. online casino market, live in six states as of early 2024 with several more in various stages of legislation, has been unusually willing to regulate slot features at the state level. New Jersey's rules on auto-play, turbo spin, and loss-limit prompts are the most prescriptive. Michigan has leaned toward disclosure requirements rather than feature caps. Pennsylvania has focused on self-exclusion integration. The result is that a game built for one state may need three different auto-play configurations, which pushes studios toward building the most restrictive version as the default and letting operators loosen it where allowed.
At the same time, the social casino and sweepstakes market, which operates outside the same regulatory perimeter in most states, has kept auto-play largely unrestricted. That creates a natural experiment. Players who move between a regulated real-money app and a sweepstakes app see two very different auto-play experiences, and the sweepstakes version, with its 100-spin and 500-spin options, is the one that matches the 61% preference. Whether that preference follows players back into the regulated market is an open question, but the direction of the pressure is obvious.
There is also the land-based comparison, which gets less attention than it should. A slot machine in a Pennsylvania casino does not have auto-play in the online sense; it has a spin button and a human finger. The 25-spin cap in the online market is, in effect, an attempt to import some of that physical constraint into a medium where the constraint does not exist. It is a reasonable instinct. It is also fighting a preference that the medium itself created.
What operators actually do with the data
Ask a product manager at a U.S. online casino about auto-play and you will usually get one of two answers. The first is that auto-play is a retention feature with a compliance wrapper, and the spin cap is a cost of doing business. The second is that auto-play is a responsible-gambling feature that happens to be popular, and the cap is a genuine safeguard.
Both are partly true, which is why the policy conversation tends to stall. The 61% figure is inconvenient for the first answer because it suggests the cap is not doing much. It is inconvenient for the second because it suggests players are routing around the safeguard by design, not by accident.
The operators that have handled this best tend to do three things. They make the loss limit the first field on the panel, not the last. They show a running total of the session's net result on the auto-play screen itself, not buried in a menu. And they treat a completed auto-play session as a natural checkpoint, prompting a re-confirmation rather than letting the next session start automatically. None of these are dramatic. All of them are cheap. The ones that skip them usually cite the same reason: the A/B tests show a revenue dip, and the responsible-gambling team does not have the authority to override it.
That is the honest version of how these decisions get made, and it is worth stating plainly because the alternative is a press release about "player-first design" that does not survive contact with the session logs.
The measurement problem underneath all of this
There is a deeper issue that the 61% figure exposes. The industry does not have a shared definition of what auto-play usage means. One operator counts activations. Another counts spins. A third counts sessions where auto-play was used for more than half the total spins. A fourth counts only auto-play on games above a volatility threshold. The numbers are not comparable, which means the responsible-gambling research literature on auto-play is thinner than it looks.
The 25-spin cap makes this worse, because it fragments a single player behavior into multiple recorded events. A player who wants 100 spins and is capped at 25 produces four activations, which inflates the activation count and makes the cap look like it is being used more than it is. Any researcher comparing operators on "auto-play usage" is comparing four different measurement conventions and calling it a trend.
The fix is not complicated: publish the definition alongside the number. Very few do. Until that changes, the 61% figure should be read as a signal about player preference, not a precise measurement of it, and the 25-spin cap should be read as a policy choice that has never been validated against the behavior it was meant to shape.
The open question is whether the cap should be redesigned around the preference it is clearly failing to change, or whether the preference itself is the thing that needs to be understood better before anyone touches the cap again. A 100-spin auto-play with a well-placed loss limit and a re-confirmation prompt might do more for player protection than a 25-spin cap that players simply run four times in a row. Nobody has run that comparison at scale in a U.S. regulated market yet. The operator that does will have something more useful than a compliance checkbox to show for it.