RTP and player payback: why transparency matters

Return to player is the most quoted number in online casino games and one of the least understood. A slot listed at 96% does not hand back £96 for every £100 a given player stakes. The figure is a theoretical, long-run average built into the game's maths. That makes it useful, but only if players know which figure applies to the game in front of them, and that is getting harder.

What the number measures

RTP is the share of all stakes a game is designed to pay back as winnings over its expected lifetime. The rest is the house edge. A 96% game has a theoretical edge of 4%. A 92% version of the same game has twice that.

Because RTP is an average over a very large number of plays, it says nothing about volatility or about any single session. Two games with the same RTP can play completely differently. That is why regulators treat RTP as one piece of information about the chances of winning, not as a promise.

How RTP is tested and monitored

Labs check RTP mostly on paper and in simulation. They review the game's maths and paytables, confirm the theoretical return across wagering options and test the random number generator. GLI-19, GLI's widely used standard for interactive gaming systems, sets a default floor: a house-banked game must theoretically pay out at least 75% over its expected lifetime unless the regulator sets a different figure, and that minimum must hold for every wagering configuration.

The same standard expects operators to keep documentation of each game's theoretical RTP and to record every change to it. A game whose theoretical RTP changes is treated as a new game for reporting. Operators should also compare theoretical and actual RTP regularly and investigate large gaps.

Great Britain goes further on live monitoring. The Gambling Commission expects licensees to calculate actual RTP and compare it with the expected, advertised figure, at a frequency that reflects how much a game is played, and to watch for both underpayment and overpayment. Its guidance is blunt on one point: telling players that a game met the testing standards is not enough, because errors can slip past testing. The Commission's technical standards also require outcomes to be "acceptably random" and ban compensated games, meaning games that change the likelihood of winning based on previous payouts or takings.

One game, several RTPs

Many games now exist in several certified RTP versions, and the operator chooses which one to run. iGaming Business reported in April 2026 that operators can deploy different RTP versions of the same game, typically 96%, 94% or 92%. GLI-19 allows for this kind of choice in general terms: systems can have configurable features whose final setting is chosen by the operator, and the configuration used in production must be passed to the lab.

There is commercial pressure behind the trend. In the UK, Remote Gaming Duty rose from 21% to 40% of profits from 1 April 2026. In the same iGaming Business report, Helen Walton, co-founder of studio G Games, described RTP as a pricing lever operators were already pulling, with standard settings moving from around 96% to 94%.

Running a certified lower-RTP version is not a breach in itself, as long as it is disclosed the way the rules require. The real question is whether a player can tell which version they are playing.

What the rules ask for

Great Britain is a useful reference because its requirements are public. Under RTS 3C of the Commission's remote technical standards, information that lets a customer make an informed decision about their chances of winning must be easily available before they gamble. That can be the house edge, the RTP percentage or the probability of winning events, and the guidance lists the average theoretical RTP as one acceptable form. For jackpot games, the RTP can be shown as one combined figure or split between the base game and the jackpot.

GLI-19 leaves the decision to show RTP to each regulator. Where RTP is shown, the game must explain how the figure was worked out: minimum, maximum or average. If an actual rather than a theoretical figure is shown, the game must also give the number of plays and the period behind it.

Where disclosure falls short

The weak points are mostly practical. The RTP usually sits in a help screen that most players never open. The RTP version is chosen site by site, so two casinos can run the same title at different settings, and a review quoting the "official" RTP may be quoting a version a particular casino does not run. And players have no easy way to check that the number in the help file matches the certified build on the server. In the UK, the lab report goes to the regulator, not to the public.

"Player payback" also has a second meaning in this industry: how quickly a player acquired through marketing earns back what it cost to acquire them. A case study on player lifetime value versus CPA payouts published on affhouse.org shows the operator side of that sum: 365 depositors bought at a flat $25 CPA generated about $168 each in gross gaming revenue, according to self-reported figures the article notes were not independently verified. In both meanings, the economics depend on numbers that outsiders usually cannot check.

What better transparency looks like

Better RTP disclosure does not need new maths. It needs the live configuration shown inside the game, ideally next to the lab and certificate reference for that exact version. It needs a record whenever that configuration changes. And where operators already compare theoretical and actual RTP, it means publishing that comparison over a stated number of plays and a stated period, which GLI-19 already expects whenever an actual RTP is displayed.

None of this changes the house edge. It changes who can see it, and in a market where the same game can carry different returns on different sites, that is the part players and partners are missing.