Finding one: 71% of self-described intermediate players admitted to at least one scoring belief that contradicts the published rules of the format they were playing. The most common contradiction was around joker rules — players who thought jokers could complete a pure sequence, or who thought two jokers per meld were universally permitted.
Finding two: 64% of players had never read the operator's published rules before their first deposit. The most common reason given was that the rules were assumed to be "standard 13-card rummy" — which is true only when the operator specifically publishes a standard-compliant ruleset. The minority who read the rules before the first deposit reported a 22% lower rate of surprise at the table.
Finding three: 58% of players had never set a session stake limit before sitting down. Players who had set a limit reported a 31% lower median loss per session than those who had not, even when controlling for stake level and format. The correlation holds across formats, which suggests the limit is doing the work — not the format's lower variance.
Finding four: 41% of players reported playing through at least one session that they had retrospectively decided to end early. The most common reason for the late-end decision was "the next hand felt lucky". The next hand was not lucky. This is the tilt signal that the responsible-play page addresses directly.
Finding five: 23% of players reported having filed a support ticket about a withdrawal that took longer than the published processing time. Of those, 67% reported that the resolution arrived within the next 48 hours. The 33% who did not get resolution within 48 hours were the readers who most needed the customer-care guidance we publish.
The survey was self-selected (respondents chose to participate), not random. Self-selected samples skew toward more engaged players, which probably understates the rate of mistakes in the broader population. We name the methodology on the operator reference page; readers who want to re-run the survey on their own sample can do so.