A bonus should not go to a player who was going to deposit anyway. That sounds obvious, and it is rarely how bonus budgets are spent. The same offer sent to a whole segment reaches the player who needed a push, the player who needed nothing, and the player who will take the bonus and leave. Here is how to tell them apart, what to do for each, and how to check afterwards what the bonus really did.
Why is one bonus for the whole base expensive?
Because the cost is paid on every recipient, and the benefit comes only from the ones who changed their behaviour. If half of a segment would have deposited without the offer, you have paid for a result you already had. The honest measure of a bonus is not how many deposits followed it. It is how many deposits followed it that would not have happened otherwise.
Retention budgets also tend to be spread evenly because it is simple. Sending one offer per cohort means the high-value player and a random one get the same message. The result is a budget that follows convenience, not value.
Which players should usually be left out?
Three groups come up again and again:
- Players who are winning right now. Someone enjoying a good run does not need an incentive to come back tomorrow. In the Sellrise demo, a player marked as winning right now is shown with the note that no bonus is needed, and the contact is moved to the next day.
- Players whose rhythm says they will return soon. A regular who deposits every Friday and has not yet reached this Friday is not lapsed.
- Players whose behaviour suggests bonus extraction. Offering more to players who play only to take a bonus feeds the problem. See casino bonus abuse and how to spot it.
A fourth category belongs in a different article, but it matters most: self-excluded players and those flagged for responsible gaming must never receive promotional contact. That is a gate, not an optimisation. See responsible gaming and casino retention.
How do you decide, player by player?
Combine three things: how likely the player is to leave or to deposit without help, what the player is worth, and what stage they are at. A new player waiting for a second deposit, a cooling-off regular and a lapsed VIP each need something different, and sometimes nothing.
This is the idea behind bonus matching in Sellrise: a promotion from your catalogue matched to the player's stage and value, with the reason shown, plus the option to say do not give a bonus now. The reason matters. A manager who sees why a player was left out can overrule it. A manager who sees a list with no explanation will ignore it.
You can do a rough version by hand: sort your active players by recent trend in bets and deposits, and hold the offer for those whose trend is rising. It is imperfect, but it is closer to the principle than a blanket send.
What do you do for the players you leave out?
Not necessarily nothing. A message without a bonus costs little: a reminder of a new game, a tournament date in your product, or just a quiet period. For players you are certain to keep, the sensible use of the budget is often to save it, and for players who are near the edge, to move your attention to another day when the signal has changed.
What matters is that leaving a player out is a decision with a reason, and that you can see later whether it was right.
How do you check what a bonus actually brought?
Compare the players who received it with a similar group who did not. The comparison needs a control group that nobody contacts, held aside from the start. Without it, the deposits after the bonus are simply the deposits after the bonus.
One of the things Sellrise shows is what bonuses actually did: deposits after the bonus, and how much of that the bonus itself brought. The second number is the one worth looking at, and it needs the control group to exist. I describe the full method in how to measure whether a casino bonus worked.
What are the risks of leaving players out?
Two. First, the forecast can be wrong. A player you decided did not need a bonus leaves. That is why you test on a held-out group before changing the whole programme, and why I would run the first month with your team deciding every send. Sellrise suggests, a person decides, and nothing goes out on its own.
Second, a policy of withholding bonuses can change player behaviour over time. If a segment is used to monthly offers, removing them may look like a loss in the short run. Watch average turnover per player, not just conversion on the offer.
Where do you start?
Pick one campaign you run regularly. Hold back a random group from it. After a month compare turnover. If the held-back group did nearly as well, you have found money you do not need to spend. If it did clearly worse, the campaign is doing its job.
If you want the per-player version, with the reason for every decision, the page on offers per player explains the approach and the demo shows it on synthetic data. You can also start a pilot and measure it on your own base.