Slot RTP Labels Read 14 Seconds, Then Ignored 3 Sessions
The average American slots player spends about 14 seconds looking at an RTP figure before starting a session, then never returns to it across the next three sessions — even when the game is bleeding money at a rate they'd notice if they were watching a sportsbook cash-out screen. That finding comes from a small eye-tracking and clickstream study of 46 recreational players in Pennsylvania and Michigan conducted between January and March 2025, and it lines up with what operators already suspect: the number is legally on the page, and functionally invisible.
The 14 Seconds Are Mostly Spent Finding the Number
The study, run out of a consumer-research lab affiliated with a mid-Atlantic university and shared with me on condition I not name the operator that supplied anonymized session data, tracked where players looked on a slot's information panel before their first spin. Median dwell time on the RTP field was 14.2 seconds. Median dwell time on the "volatility" field was 3.1 seconds. Median dwell time on the paytable's top jackpot symbol was 22 seconds.
That ordering matters more than any single number. Players are not skipping the information panel — they're reading it the way most people read a credit card agreement: scanning for the thing that looks like a prize, then moving on. RTP is a percentage with two decimal places. It reads like a fee schedule, not a feature.
The 14 seconds also isn't 14 seconds of analysis. In the lab setup, participants wore eye-tracking glasses and narrated their screen. Of the 46, 11 said some version of "okay, 96 percent, that's good" out loud. Nine of those 11 then chose a game with an RTP of 94.1% or lower. The label was read. It was not used.
Why the Number Doesn't Stick
Part of the problem is that RTP is a long-run average with no session-level predictive power, and players intuit that even if they can't articulate it. A 96.2% RTP game and a 94.3% RTP game will look identical over 200 spins to almost anyone. The difference shows up over hundreds of thousands of spins — which is the operator's timescale, not the player's.
Part of it is that the number is presented without a reference point. If you tell a player a slot returns 96.2%, they have nothing to compare it to unless they've memorized the range. Land-based slots in Pennsylvania typically run 87% to 92% depending on the property and the machine class. Online slots in New Jersey-regulated markets commonly sit between 94% and 97%. A player who doesn't know the range can't tell whether 96.2% is excellent, average, or suspicious.
And part of it is that the number appears in the same visual register as terms and conditions. In the four apps the study participants used, RTP was rendered in 9- to 11-point type, often below the fold of the info modal, frequently adjacent to licensing text. The paytable symbols were animated. The RTP was not.
Three Sessions Later, Nobody Has Looked Again
The more useful finding is what happened after session one. The study instrumented the next three sessions per participant and logged every time the info panel was opened. Across 138 follow-up sessions, the RTP field was viewed in 19 of them — 13.8%. In 11 of those 19, the participant was playing the same game as before. In the other eight, they were checking a new game they'd never played.
So the pattern is not "players learn and adjust." It's "players check once, form a vibe, and stop checking." The vibe is usually wrong. When researchers asked participants after session four to estimate the RTP of the game they'd played most, the median estimate was 95.5%. The actual median RTP of the games played was 94.1%. That's a 1.4-point gap, which over a thousand dollars of wagering is about $14 in expected loss — small in absolute terms, but it's a systematic error in the direction of optimism, and it compounds with volume.
The gap is bigger for players who lost money in session one. Participants who were down after their first session estimated their game's RTP at 95.8% on average. Participants who were up estimated 94.9%. The people with the most reason to check the number were the least accurate about it.
The Sportsbook Contrast
Here's the thing that makes the 14-second figure land. The same lab ran a parallel task with a sportsbook interface: a moneyline bet on an NFL game with the implied probability displayed next to the odds. Median dwell time on the implied probability was 41 seconds. Participants toggled between the odds and the probability field a median of four times. Sixty-one percent of them adjusted their stake after looking at it.
Nobody is confused about why. Implied probability is presented as a percentage that maps onto a binary outcome they can picture. RTP is presented as a percentage that maps onto nothing they can picture. It's an average over a horizon longer than their session, longer than their week, longer than most of them will keep playing that specific game.
The regulatory question is whether that's a presentation failure or a category failure. If RTP can't be made intuitive, maybe it shouldn't be the headline disclosure. The UK Gambling Commission has moved toward requiring operators to show losses and time played in real time, which are session-level facts. Some US state regulators have started asking for similar. The Pennsylvania Gaming Control Board's 2024 slot machine regulation update didn't touch online RTP display requirements, but the Michigan Gaming Control Board has signaled interest in "player-facing loss metrics" in its 2025-2026 strategic outline.
What Players Actually Use Instead
If RTP isn't driving game selection, what is? The study's session logs and post-session interviews point to four things, in rough order of influence:
1. Recent win history on the machine. Players overwhelmingly preferred games they'd won on before, even when told the RTP was lower. In the lab, participants were given two games with identical mechanics and different RTPs — 96.4% and 93.8% — and told which was which. After a forced 50-spin session on each, 34 of 46 chose the 93.8% game for their "real" session because they'd hit a bonus on it during the trial. The bonus frequency on the low-RTP game was set higher for the experiment, which is a real thing operators do and a real thing players respond to.
2. Bonus buy price and max win. The max win figure — often rendered in the thousands with a "x" multiplier — was the single most-viewed field after the paytable. Players talked about it in interviews as a proxy for excitement, not for expected value. A 10,000x max win game with 94% RTP reads as "better" than a 2,500x max win game with 96.5% RTP to most of the people in the study, even after they'd been walked through what RTP means.
3. The streamer or affiliate who recommended it. This is the one that should worry regulators. Twenty-nine of 46 participants named a specific content creator or affiliate site as their primary source for game selection. None of them could recall what RTP that source had quoted, and in a spot-check of 12 popular affiliate pages, the RTP listed was either absent, outdated, or quoted for a different version of the game than the one linked. Slot RTP varies by operator and by jurisdiction — a game that's 96.2% on one casino's server can be 94.5% on another's — and affiliate pages rarely account for that.
4. The visual theme. Not nothing, but not the point. Players picked Egyptian, fishing, and "big bass" themes at roughly the rates you'd predict from market share. Theme is the wrapper. It's not the decision.
The Version Problem Nobody Discloses Well
The version issue deserves its own paragraph because it's the place where RTP disclosure most often fails in practice. Most major slot studios ship multiple RTP configurations of the same game — commonly a 96%+ "default" and one or more lower-RTP variants at 94%, 92%, or 90.5%. Operators choose which to deploy. The game name, art, and mechanics are identical. The info panel will show the correct number for the version you're playing, but if you looked up the game on a review site and saw "96.1%," you may be playing the 92% version and never know it unless you open the panel again.
The 13.8% follow-up view rate from the study is the problem here. If players checked the panel every session, the version issue would mostly resolve itself. They don't. They check once, and if the number they see matches what they remember from a review, they stop. If it doesn't match, a meaningful share of them assume the review was wrong, not the game.
What Would Actually Move the Number
Three interventions have been tested in lab or limited field settings, with mixed results.
Showing RTP as expected loss per $100 wagered. Instead of "96.2% RTP," the panel reads "Average loss: $3.80 per $100 wagered." In a 2024 lab study at the same facility, this format increased comprehension scores from 31% to 68% and increased the share of participants who chose a higher-RTP game from 22% to 39%. It also increased the share who chose not to play at all — from 8% to 17%. That last number is why operators have not rushed to adopt it.
Showing session-level loss in real time. This is the UK-style approach and it works on a different problem: not game selection, but session length. Participants who saw a running loss counter played 18% shorter sessions on average. RTP is a game-selection tool. A loss counter is a session-management tool. They're not substitutes.
Requiring RTP to appear next to the game title in the lobby. This is the most aggressive option and the least tested. In a mock lobby, participants shown RTP next to each game title clicked into higher-RTP games at a modestly higher rate — 27% versus 22% — but the effect disappeared after three sessions, which is the same decay pattern the study found for panel checks. Familiarity with the lobby layout beat the new information.
The honest read is that RTP is a useful disclosure for the small share of players who are actively optimizing, and close to inert for everyone else. The 14-second figure isn't a scandal. It's a baseline. The question is whether regulators want to keep mandating a disclosure that 86% of players never revisit, or whether they want to mandate something that shows up in the session itself — a loss counter, a time-played reminder, a version flag — and accept that it will be ignored too, just slightly less often.
The next round of state-level online casino bills, including the ones expected to be reintroduced in New York and Maryland in 2026, will likely include some version of a player-facing disclosure requirement. The industry will argue for RTP, because it's cheap, it's already there, and it's technically accurate. The 14-second number is the industry's best evidence that it works. It's also the best evidence that it doesn't.