A 60/40 split between bonus credit and cashable balance reduced the average depth of session-90 rebuys by 14.2% across a 4,318-player cohort tracked between January and June 2024. The effect held at 13.6% after controlling for stake size, deposit frequency, and game category, which places it outside the noise band the operator's own A/B framework uses to flag false positives. The mechanism is not the bonus itself but the sequencing: players who cannot convert promotional credit into withdrawable funds until a defined trigger are measurably less likely to chase losses past their ninth session.
The Structure Being Tested
The operator in question — a mid-tier US-facing brand with a New Jersey license and roughly 190,000 monthly active accounts — ran the split as a modification to its standard reload offer. Under the control condition, a $50 reload produced $50 in bonus credit with a 25x wagering requirement applied to the bonus alone. Under the treatment condition, the same $50 reload produced $30 in bonus credit and $20 in cashable balance, with the wagering requirement applied only to the $30 portion.
The headline difference is smaller than it looks. Total promotional value is identical at $50. The wagering burden is higher in relative terms under the treatment (25x on $30 rather than 25x on $50), but the absolute dollar exposure required to clear is 40% lower. That asymmetry is the entire intervention. Players in the treatment arm face a smaller wall to climb, but they also have $20 that is already theirs.
Session-90 rebuy depth is defined here as the cumulative deposit amount made by a player between their 85th and 95th session, expressed as a multiple of their median session-1 through session-10 deposit. It is an awkward metric, and the operator's analytics team acknowledges as much. But it captures something a simpler measure — total deposits, or churn rate — does not: the degree to which a player escalates their financial commitment as their account ages.
Why the 14% Figure Matters More Than It Sounds
A 14.2% reduction in rebuy depth does not mean players deposit 14% less. It means the rate of escalation slows. In the control arm, the median player at session 90 was depositing 3.4x their early-session median. In the treatment arm, that figure was 2.9x. Both cohorts are still escalating. The treatment arm is simply escalating more slowly, and the gap widens with session count — at session 150 the divergence reaches 19.7%, though the sample thins considerably by that point and the confidence interval widens to roughly ±6 points.
The operator's responsible gambling team flagged the result as directionally useful but cautioned against treating it as a harm-reduction intervention. The 14% figure sits within the range of what the team calls "structural friction effects" — changes that alter behavior without changing underlying risk profile. A player who deposits less per session but plays more sessions may end up in the same place. The six-month window is not long enough to say.
Where the Effect Disappears
The split did not perform uniformly. Three subgroups showed no statistically significant change in session-90 rebuy depth, and their composition is instructive.
High-frequency depositors. Players making more than 12 deposits per month in their first 30 days showed a 3.1% reduction, well inside the noise band. This group tends to treat promotional credit as a rounding error against their baseline activity. A $20 cashable component is not a meaningful change to their bankroll structure.
Single-game-category players. Players who confined more than 90% of their wagering to one vertical — most commonly slots — showed a 5.8% reduction. The split appears to work best on players who move between sportsbook and casino, where the cashable portion has immediate utility in a way it does not for a slots-only player who is going to re-deposit into the same game regardless.
Players with prior self-imposed limits. This group showed a 1.2% increase in rebuy depth, which is almost certainly noise but is worth flagging. Players who have already set deposit limits may be less responsive to bonus structure changes because their behavior is already constrained by a harder boundary.
The implication is uncomfortable for operators hoping to deploy this as a blanket policy. The 14.2% headline is a weighted average across a cohort where roughly 40% of players showed no meaningful response.
The Confound Nobody Has Resolved
The control and treatment arms were not identical in one respect the operator did not initially flag: the treatment arm's cashable component was available for withdrawal immediately, subject to the standard 1x playthrough on the deposit itself. This means treatment-arm players could, in principle, deposit $50, receive $30 bonus and $20 cash, wager $20 once, and withdraw $20 — a net loss of $30 against a $50 deposit, which is a worse outcome than simply not depositing.
Very few players did this. The rate was 2.3% in the treatment arm versus 0.4% in control. But those players are not distributed randomly; they cluster in the first 14 days of account life, and they churn at a higher rate than either arm's median. Whether the split caused their churn or merely accelerated a churn that was already coming is not answerable from this dataset.
What the Numbers Do Not Say
The operator's report is careful to note that rebuy depth is a proxy, not an outcome. It correlates with problem gambling severity in the literature — the PGSI items around escalation and chasing are the relevant ones — but correlation at the cohort level does not license inference at the individual level. A 14% reduction in a median does not mean 14% of players were helped.
There is also the question of what happens when the cashable portion is removed. The operator ran a third arm — $50 bonus credit, no cashable component, 25x wagering — and the session-90 rebuy depth was 1.8% higher than control. That is within noise, but it suggests the effect is not simply "less bonus is better." The cashable component appears to be doing work that the bonus reduction alone does not.
If that holds, the policy implication is not "reduce bonus sizes." It is "make part of the offer immediately real." Those are different interventions with different cost structures, and only one of them is likely to survive contact with a marketing department that measures success in conversion rate rather than session-90 behavior.
The open question is whether a 14% reduction in escalation depth over six months translates into any measurable change in harm at 24 months, or whether it simply reshapes the curve without bending it. The operator has committed to a 36-month follow-up. Until that data exists, the honest position is that this is a promising structural finding with an unresolved mechanism and an unknown terminal effect.