Slots Lobby Sort by "Newest" Cuts Session Length 14%
A default sort order is usually treated as a neutral piece of interface plumbing. It isn't. In a controlled test run across three U.S.-facing casino lobbies between January and March 2025, moving the default landing sort from "Popular" to "Newest" cut median session length by 14.2% — from 31 minutes 40 seconds to 27 minutes 10 seconds — while leaving deposit conversion essentially flat at a 0.4-point decline. The finding matters because session length is one of the few metrics that both operators and responsible-gambling researchers watch, and they watch it for opposite reasons.
The test wasn't run by an operator. It was run by a mid-size platform vendor that supplies lobby and search infrastructure to eight licensed U.S. brands, which shared anonymized, aggregated event data on the condition its name and its clients' names stay out of print. That's a limitation worth stating up front: this is vendor-side telemetry, not a peer-reviewed study, and the vendor has a commercial interest in selling lobby optimization tools. But the sample is large — 412,000 unique accounts, roughly 9.1 million sessions, split 50/50 between a control cohort that kept "Popular" as the default and a treatment cohort that landed on "Newest" — and the effect size is stable enough across the three brands to be worth taking seriously.
What "Newest" actually changes
The mechanism is less mysterious than the number suggests. A "Popular" sort is a feedback loop. It surfaces the games that already have the most traffic, which generates more traffic, which keeps them surfaced. On a typical lobby, the top 20 slots by popularity account for somewhere between 55% and 70% of all slot spins, even when the catalog holds 800 to 1,500 titles. That concentration is not a secret; every slot provider knows its revenue is carried by a handful of evergreen titles and a rotating cast of new releases.
A "Newest" default breaks the loop. It puts fresh releases in front of players who did not ask for them, and fresh releases behave differently in two measurable ways.
First, they have shorter average play arcs. The vendor's data shows a new-title session — defined as a session where the player's first spin is on a game released in the prior 60 days — runs a median 11.3 minutes, against 19.8 minutes for a session that opens on a title older than 18 months. Some of that is novelty seeking: players sample, form an opinion, and leave. Some of it is mechanical. New releases increasingly ship with lower default bet sizes in the lobby's quick-play mode, and a player betting $0.40 a spin hits their personal loss threshold faster than one betting $1.00.
Second, "Newest" scatters attention. In the control cohort, 38% of sessions involved a single game and no lobby return. In the treatment cohort, that figure fell to 29%, but the number of distinct games touched per session rose from 1.7 to 2.9. Players were browsing more and committing less. Total spin count per session dropped 9.6%, and total wagered per session dropped 12.1% — slightly less than the drop in time, meaning the remaining time was spent at marginally higher average stakes.
That last detail is the one that complicates the story. If you only read the headline — sessions 14% shorter — you'd assume the operator lost money. It didn't, at least not in this test. Gross gaming revenue per session fell 11.4%, but sessions per user per week rose 6.8%, and 30-day retention for the treatment cohort was 2.1 points higher. The vendor's own summary is that "Newest" trades depth for frequency. Whether that's a good trade depends entirely on what you're optimizing for, and that's where the responsible-gambling angle stops being a footnote.
The two audiences reading the same number
For an operator, a 14% shorter session is a cost. For a public-health researcher, it's a benefit. Both are reading the same dashboard.
The academic literature on gambling harm has converged on session length and session frequency as two of the more reliable behavioral markers associated with problems, though neither is diagnostic on its own. The standard instrument in U.S. research is the Problem Gambling Severity Index, and the correlation between PGSI scores and time-on-site is well established but modest — typically in the 0.2 to 0.35 range depending on the study and how session boundaries are defined. What matters more is the interaction: long sessions, high frequency, and escalating stake size together predict harm far better than any one variable.
The vendor's treatment cohort moved in a mixed direction on exactly those three variables. Sessions got shorter. Frequency went up. Average stake per spin went up slightly. A harm-reduction researcher looking at that pattern would not call it an unambiguous win. Shorter sessions are generally considered protective, but more frequent returns to the product, with slightly higher stakes, is not a profile anyone should describe as safer without more data.
There's a second-order effect the test didn't measure and probably couldn't. "Newest" as a default means the lobby's most prominent real estate goes to titles with no track record. New slots are not systematically higher-variance than older ones — variance is a property of the math model, not the release date — but they are systematically less familiar to players. A player who knows a game's bonus cadence and dead-spin patterns is playing a game whose rhythm they can predict. A player sampling a title for the first time is, in a narrow but real sense, playing with less information. That doesn't make new games more dangerous. It makes them less legible, and legibility is part of what lets a player decide to stop.
Where the 14% comes from, and where it doesn't
The number is a median, and medians hide the shape of a distribution. When the vendor broke sessions into deciles, the compression wasn't uniform.
| Session decile (control) | Median length, control | Median length, treatment | Change |
|---|---|---|---|
| 1st (shortest 10%) | 4m 10s | 3m 50s | −8.0% |
| 3rd | 12m 30s | 10m 55s | −12.7% |
| 5th (median) | 31m 40s | 27m 10s | −14.2% |
| 7th | 58m 20s | 47m 05s | −19.3% |
| 9th | 2h 14m | 1h 39m | −26.1% |
The effect is strongest at the tail. The longest 20% of sessions — the ones that account for a disproportionate share of total wagered and, not coincidentally, the ones that most resemble the behavioral profile associated with harm — shrank by roughly a quarter. The shortest sessions barely moved. If you were designing an intervention specifically to trim the long tail of extended play, a default sort change would be a strange but not ineffective way to do it.
That's also the finding's biggest caveat. The tail shrank, but it didn't disappear. The 9th decile still had sessions running over an hour and a half. And the test period was 11 weeks. Behavioral adaptation to interface changes is real and usually runs in the direction of the effect decaying: players learn to navigate around a default they don't like. The vendor's own follow-up data at week 11 showed the treatment cohort's session length had drifted back up to 28m 40s, narrowing the gap from 14.2% to 9.5%. Extrapolate that curve and the effect may be closer to 5% by month six.
What operators are actually doing with this
Three of the eight brands on the vendor's platform have since made "Newest" the default for logged-out and first-session users, then reverted to "Popular" or a personalized sort once an account has 30 days of history. That hybrid is a reasonable reading of the data: capture the novelty-seeking behavior of new players, then let established players return to the games they actually want.
One brand went the other direction and made "Newest" the default for everyone, paired with a soft prompt after 45 minutes of continuous play. Its reported 90-day numbers — which I could not independently verify — show session length down 11% and deposits per user up 3%. If accurate, that's a genuinely awkward result: the intervention that shortens sessions also increased revenue. Harm-reduction advocates have long argued that shorter sessions and healthier revenue are not mutually exclusive, and this is a small piece of evidence for that position. It's also exactly the kind of result that gets over-claimed by vendors selling the tool.
The more interesting question is whether any of this is stable enough to build policy on. Most U.S. regulated markets impose no interface-level requirements. The rules that exist tend to be about disclosures, self-exclusion tools, and advertising, not about default sort orders or lobby composition. A 14% session-length effect from a dropdown menu is the kind of finding that makes regulators ask whether default settings should be in scope. It's also the kind of finding that's easy to poke holes in: single vendor, self-reported, no pre-registration, no control for game mix, and a follow-up trend showing the effect weakening.
The honest read is that this is a signal, not a settled result. It suggests that lobby architecture is a behavioral lever with real magnitude — larger than most operators assume when they A/B test button colors and call it optimization. It does not yet tell us whether pulling that lever makes players better or worse off, because the metric everyone reached for first, session length, is only one of three that matter, and the other two moved the wrong way.
What would settle it is a study that tracks PGSI scores, not just session timestamps, across a sort-order change — and that runs long enough to see whether the effect decays to zero. Nobody in this test did that. The vendor has the data to attempt it and a commercial reason not to publish anything that makes its product look like a harm-reduction tool one quarter and a revenue tool the next. Until someone runs that study, the 14% figure will keep getting cited by people on both sides of the argument, each reading it as proof of their own position.