Use NGR to judge retention, and keep GGR for context. Gross gaming revenue is total bets minus total wins, before bonuses and costs. Net gaming revenue subtracts bonuses, provider fees, platform and payment costs. A retention campaign spends bonus money by definition, so a metric that ignores bonuses will make almost any campaign look good. I will explain the difference, where each number belongs, and what it means when someone proposes a fee as a share of GGR.
What is the difference between GGR and NGR?
Our glossary puts it plainly. GGR, or gross gaming revenue, is all bets minus all wins, before bonuses and costs. NGR, or net gaming revenue, is GGR minus bonuses, provider fees, platform and payment costs.
The gap between the two is where retention lives. Bonuses are the main lever of a retention team, and they sit between GGR and NGR. A team that reports GGR can double its bonus spend and still show growth, because the cost never appears in the number it reports.
Which one should you use to judge a campaign?
Whichever includes the cost of the campaign. In practice that means NGR, or a metric built from it. If you use turnover, which is the metric we use in a Sellrise pilot (average turnover per player), then also record the cost of the bonuses and channels used, so the result can be read net of what it took to get there.
Turnover has one advantage as a test metric: it is simple and hard to dispute. But it does not tell you whether you made money. The two views belong together: turnover to see whether behaviour changed, and net revenue to see whether it was worth it. I cover the test design in how to run a control group test.
When is GGR the right number?
- Comparing the size of products or markets, where you want volume before costs.
- Reading the market and regulatory reports, where GGR is the common unit.
- Contract terms. Some agreements are tied to it, which brings me to the next point.
GGR is not a wrong number. It is a number for a different question. The mistake is using it to answer whether a retention action paid for itself.
What does it mean when a vendor's fee is a share of GGR?
It means the vendor earns more when your gross revenue rises, including growth that has nothing to do with them, and including growth bought with bonuses. I discuss the trade-offs in how casino CRM is priced. A revenue share is not automatically bad. It lowers upfront risk. But you should ask which revenue it applies to, how attribution is measured, and whether a control group is used.
Sellrise is a subscription by volume and not a share of GGR. I am biased and I say so. The point I would hold to whatever you choose is that the metric your vendor is paid on should not be easier to inflate than the metric you care about.
How do bonuses distort the picture at player level?
A player can look valuable in GGR and poor in NGR. A bonus hunter who deposits, plays through a wagering requirement and withdraws may leave a gross figure that looks fine and a net figure near zero or below. If your VIP process ranks by gross, it will pick the wrong people. See how to spot future VIP players in their first week.
Likewise, a player who receives frequent offers can show healthy gross revenue while costing a lot to keep. Seeing the net per player is the only way to find that out. It is also why Sellrise tries to show what bonuses actually did: deposits after the bonus, and how much of that the bonus itself brought.
What cost data do you need to compute NGR per player?
Bonuses granted and redeemed are the key piece, and they are in the deposits and bonus records of most platforms. Provider fees, platform costs and payment costs are often only available as totals, and allocating them to individual players needs rules that you and your finance team agree on.
I would not wait for a perfect allocation. Start with bonuses, which are the part retention controls, and add the other costs as the data allows. A rough NGR per player that includes bonuses is more useful for retention than a precise GGR that does not.
How does this affect the forecasts?
A forecast is only as useful as the value it predicts. If the value is gross, it will recommend contacting players whose net is poor. In Sellrise the forecast gives value over 90 days as a range, and the economics of a bonus (margins, wagering terms, legal limits) stay on your side, since a model cannot know them. That separation is on purpose. I read the model's job as saying who and when, and yours as saying how much it is worth to you.
Read what predicted lifetime value means for how the range should be used.
What should you do next?
Check which number your retention reports use today. If it is GGR, add bonus cost to the same table. Pick one campaign and compute net revenue for contacted and held-back players. If you want to see how this looks inside a retention CRM, the demo runs on synthetic data, and a pilot measures it on your players.