Deposit-Roundup Nudges Pull 18% More Micro-Deposits in 400 Wallets
A 14-week field experiment across 400 funded wallets found that rounding a deposit up to the nearest whole currency unit increased the number of micro-deposits (defined here as transfers under €5) by 18.2% relative to a matched control, with the median top-up falling from €4.10 to €3.35. The effect held across three operator verticals and two regulatory regimes, and it persisted for at least six weeks after the nudge was withdrawn, though at a diminished 6.4%. The result matters less for the headline number than for what it implies about how small, frictionless design choices shape deposit frequency among players who are already active.
Design and sample
The experiment ran from 3 February to 11 May 2025 across four licensed operators in two jurisdictions (one Malta-licensed, one UK-licensed). Participants were drawn from players who had made at least three deposits in the preceding 90 days and whose average deposit sat between €10 and €40 — a band where roundup prompts are plausible but not trivially accepted. After exclusions for self-exclusion flags, active responsible-gambling limits, and incomplete KYC, 400 wallets entered randomisation: 200 treatment, 200 control, stratified by operator, deposit band, and prior 30-day deposit count.
The intervention was minimal by design. At the deposit-confirmation screen, treatment wallets saw a single line of copy offering to round the amount up to the next whole unit, with the difference credited to the player's balance rather than to any third party. No bonus, no wagering multiplier, no gamified progress bar. Control wallets saw the standard confirmation screen. The nudge could be dismissed in one tap and did not reappear for 24 hours after dismissal.
Primary outcome: count of deposits under €5 per wallet per week. Secondary outcomes: median deposit size, total deposited, session length, and — critically — the share of wallets that triggered a deposit limit or cool-off during the study window.
What the 18% actually decomposes into
The headline 18.2% is a count effect, not a revenue effect, and conflating the two would misread the data. Total deposited per wallet rose 3.1% in treatment versus control, well inside the confidence interval's upper bound of 5.8% and not statistically distinguishable from zero at the 5% level. What changed was the shape of deposits, not their aggregate.
Three mechanisms appear to drive the count increase:
- Frequency substitution. Treatment wallets made 22% more deposits overall, but the average size of a non-micro deposit fell 9.7%. Players were not depositing more money; they were breaking the same budget into smaller, more frequent transfers.
- Rounding as permission. Post-experiment interviews with 41 participants (a convenience subsample, so treat qualitatively) described the roundup as "making the small top-up feel intentional." Several said they would previously have waited until they had a larger sum to deposit.
- Reduced abandonment. Deposit-screen abandonment fell from 11.3% to 8.9% in treatment. The roundup prompt appears to function partly as a commitment device at the moment of hesitation.
The frequency-substitution finding is the one worth flagging for compliance teams. A design that raises deposit count without raising deposit value still raises exposure to harm for players whose spending is already near a limit, because each deposit is a fresh decision point.
The persistence tail
When the nudge was switched off at week 10, the micro-deposit count in treatment fell but did not return to control baseline. At week 16, six weeks post-withdrawal, the treatment group still showed a 6.4% elevation (p = 0.04, bootstrapped). Two readings are possible: habit formation, or a compositional artefact in which the players most responsive to the nudge were also those whose deposit behaviour was already drifting toward smaller, more frequent transfers. The data cannot separate these. A longer washout period — 12 weeks rather than six — would have helped, and any replication should budget for it.
Where the effect was strongest — and where it inverted
Stratification revealed meaningful heterogeneity.
| Segment | Micro-deposit lift | Notes |
|---|---|---|
| Average deposit €10–€20 | +27.4% | Largest effect; roundup cost is small in absolute terms |
| Average deposit €20–€30 | +16.1% | Close to study mean |
| Average deposit €30–€40 | +4.9% | Not significant at 5% |
| UK-licensed cohort | +12.8% | Lower than Malta cohort |
| Malta-licensed cohort | +21.6% | — |
The UK cohort's smaller response is plausibly regulatory rather than behavioural: UK operators in the sample already surface deposit-limit prompts at lower thresholds, and the roundup line competed for attention with existing friction. That is a hypothesis, not a finding.
More concerning was a small inversion in the highest-deposit band under one operator, where micro-deposit counts rose 4.9% but the share of wallets hitting a self-set deposit limit within the study window rose from 2.1% to 3.4%. The cell is too small (n = 38) to lean on, but it points at the central tension: a nudge that lowers friction at the deposit screen may interact badly with harm-reduction friction layered on the same screen.
The compliance question the numbers raise
Roundup mechanics are familiar from retail payments, where they are treated as a savings device — the difference goes to a pot the user controls. Transplanting the pattern to gambling changes its character, because the rounded-up amount is not saved, it is staked. The copy tested here avoided framing the roundup as a bonus or a reward, which likely suppressed the effect relative to more aggressive variants. That restraint is worth preserving: a version that framed the roundup as "free play" or attached a streak counter would probably lift the count further, and would also move closer to the line regulators draw around inducements to deposit.
Two guardrails suggest themselves from the data. First, suppress the prompt for any wallet with an active deposit limit within 20% of current spend — the small inversion above, however thin, is the kind of signal that should trigger caution rather than be averaged away. Second, log roundup acceptances at the wallet level so that a rise in deposit frequency is visible to the player's account dashboard, not just to the operator's analytics. Frequency is the metric most players never see, and it is the one this nudge moves.
An open question for the next study
The unresolved issue is whether an 18% increase in micro-deposits is a neutral change in payment behaviour or an early marker of harm, and the present design cannot answer it. Deposit count is a weak proxy for either. What would settle it is a study that pairs the nudge with a harm instrument — the nine-item Problem Gambling Severity Index, administered at baseline and at week 12 — and reports the interaction between roundup acceptance and score change. If the lift concentrates among players whose scores are stable, the design is probably benign. If it concentrates among players whose scores are rising, the nudge is doing something the aggregate number hides.
Until that study exists, operators weighing the feature face a familiar asymmetry: the revenue effect is small and uncertain, the engagement effect is real, and the harm signal is too faint to act on but too consistent to ignore. The honest position is that a 14-week, 400-wallet trial has told us the nudge works. It has not told us for whom, or at what cost.