A five-question responsible-gambling quiz placed between account creation and first deposit reduced average net losses at the 60-minute mark of a player's first session by 14.2%, according to a randomized controlled trial run across four licensed operators in the second half of 2024. The effect held across sportsbook and casino products and persisted, at a diminished 6.8%, through the player's first 30 days. The finding matters less for its headline number than for what it implies about where harm reduction actually happens: not in the deposit flow, but in the ninety seconds before it.
The Design
The trial, conducted by a research consortium with data-sharing agreements from four operators holding licenses in New Jersey, Pennsylvania, and Michigan, randomized 41,700 newly registered adults into two arms. The control arm saw the operators' standard sign-up flow—identity verification, payment setup, and a terms-of-service acknowledgment that, in the aggregate, fewer than one in twelve subjects reported reading. The treatment arm saw the same flow with one insertion: a five-item quiz, untimed, with no pass/fail threshold and no consequence for wrong answers.
The quiz items were not trivia. Each presented a scenario with a numeric answer the player had to estimate: the expected loss on a $500 bankroll at a 96% RTP slot over 1,000 spins; the probability that a -110 point spread bet wins; the house edge on a double-zero roulette wheel; the effective cost of a 25x wagering requirement on a $100 bonus; and the expected number of losing sessions in 20 for a player with a 1.5% edge. After each answer, the interface displayed the correct figure and a two-sentence explanation. Median completion time was 94 seconds.
The primary outcome was net loss—deposits minus withdrawals—at the 60-minute mark of the first session, measured at the operator level and reconciled against payment processor records. Secondary outcomes included session length, deposit frequency in the first week, and self-exclusion uptake at 30 and 90 days.
Why the 60-Minute Mark
Session-60 loss was chosen as the primary endpoint because prior observational work had identified the first hour as the period of steepest loss accumulation for new accounts, particularly among players who arrive through promotional landing pages. It is also short enough to be plausibly attributable to the sign-up intervention and long enough to capture real play rather than a single spin.
What the Numbers Show
The headline result: mean net loss at 60 minutes fell from $47.30 in the control arm to $40.58 in the treatment arm, a 14.2% reduction (p < 0.01). The distribution matters as much as the mean. The treatment arm showed a 19% reduction in the share of players losing more than $200 in the first hour—from 6.1% to 4.9%—while the share of players losing nothing or finishing ahead was essentially unchanged. In other words, the quiz did not suppress play broadly. It compressed the right tail.
That distinction is the interesting one. A blanket deterrent—an intrusive modal, a mandatory cooling-off period—would be expected to reduce participation across the board, and operators have historically resisted such measures on exactly those grounds. Here, total first-session handle fell only 3.4%, well within the confidence interval of no effect, while the proportion of high-loss sessions dropped materially. The intervention appears to have changed how a subset of players behaved rather than whether they played.
Secondary outcomes were more mixed:
- Session length. Median first-session length fell 8 minutes, from 71 to 63, a difference that was statistically significant but smaller than the loss reduction would predict.
- Deposit frequency. No significant difference in the number of deposits in week one (means of 2.4 vs. 2.3).
- Self-exclusion. Uptake at 30 days was 0.9% in treatment versus 0.6% in control—a 50% relative increase on a small base. At 90 days the gap narrowed to 0.4 percentage points and lost significance.
- Quiz accuracy. Mean score was 2.1 of 5. Players who scored 0 or 1 (31% of the treatment arm) accounted for a disproportionate share of the loss reduction; among players scoring 4 or 5, the treatment effect on session-60 loss was indistinguishable from zero.
That last finding deserves emphasis. The intervention worked hardest on the players who understood it least, which is either the mechanism functioning as intended—awareness landing where it was absent—or an artifact of a novelty effect that decays. The 30-day persistence data (6.8% reduction) leans toward the former but does not settle it.
The Compliance Problem
There is a regulatory wrinkle the operators involved have been quiet about. In jurisdictions that require "plain language" disclosures, a five-item numerical quiz sits awkwardly: it is more demanding than a terms-of-service clickwrap but less standardized than the responsible-gambling messaging templates most state divisions publish. Two of the four participating operators have since deployed a modified version with three items and simplified arithmetic, and neither has published outcome data.
The American Gaming Association's responsible-gambling code, updated in 2023, encourages "evidence-based" interventions at onboarding but sets no minimum. State regulators in New Jersey and Pennsylvania have shown interest in mandatory knowledge checks for sportsbook customers, and the trial's results will likely surface in comment periods. The counterargument—that a quiz which 31% of players fail outright may function as a liability shield rather than a harm-reduction tool—is not easily dismissed.
Where the Effect Comes From
The mechanism is not fully identified, and the authors are candid about it. Three candidates are plausible:
- Expectation correction. Players who overestimate their edge or underestimate house advantage may recalibrate after seeing the arithmetic. The quiz items were chosen to target precisely those misestimations.
- Friction and delay. Ninety-four seconds of forced reflection before the first deposit may interrupt an impulsive decision path independent of what the quiz teaches. A control condition with an equally long but non-numerical delay would separate these; the trial did not include one.
- Selection and signaling. The quiz may signal that the operator takes the product seriously, altering the composition of who completes registration. Registration-to-first-deposit conversion was 2.1 percentage points lower in the treatment arm, consistent with a small deterrence effect at the margin.
A follow-up trial with a delay-only arm is reportedly funded and scheduled for early 2026. Until then, the 14.2% figure should be read as the effect of the package, not of the knowledge it imparts.
The Question Worth Asking
The practical implication is narrow but real: a sub-two-minute insertion at the point of registration produced a measurable, product-agnostic reduction in early losses without materially suppressing handle, and it cost the operators roughly $0.40 per new account to implement. That is a better cost-per-unit-of-harm-reduced than most deposit-limit prompts, which typically reach players who have already formed habits.
The open question is whether any of it survives contact with a player's second month. The decay from 14.2% to 6.8% over 30 days is consistent with both a durable but fading effect and a novelty effect in free fall. If the latter, the intervention is a speed bump on a road that resumes its grade within weeks—useful, cheap, and ultimately insufficient. If the former, the industry has been overlooking the cheapest lever it has, and the burden shifts to regulators to explain why a knowledge check at sign-up is not already standard. The 2026 delay-only trial will not fully answer that, but it will tell us whether we are measuring education or simply a well-placed pause.