The Cam Clock Report

What this actually measures

One thing has to be said before any number, because it changes how every figure should be read. We observe the platform through its public affiliate feed, and that feed returns at most 200 rooms per request, ordered by popularity. Everything here therefore describes the top of the leaderboard — the most-watched rooms — and not the platform as a whole. A performer with nine viewers is invisible to us, and always was.

That is a real limitation, and it is also the reason the data is worth anything: the busiest rooms are precisely the ones whose behaviour sets the competitive picture for everybody else. We observe around the clock, and all twenty-four hours of the day carry observations.

There is a sampling bias here that we would rather state than bury. Observations are recorded when someone loads a page on this site, so an hour with more visitors is an hour in which we notice more performers. A single performer cannot be counted twice inside one hour, which stops the busiest hours inflating any individual, but it does not stop them surfacing a wider slice of the roster. The shape of the broadcast curve below therefore carries some of our own audience's daily rhythm mixed into it, and the honest reading is "when are performers visible to a European and North American audience", not "when do performers switch on". We are moving to fixed-interval sampling to remove this; until that has run long enough to publish, treat the peaks as directionally right and the exact ratios as soft.

The broadcast clock

Each bar is the number of performer-hours we observed in that UTC hour across the whole window — the supply side of the market.

Observed broadcast hours by UTC hour · 829,756 performer-hours · 2026-07-06 to 2026-08-01

The busiest broadcast hour is 02:00 UTC and the quietest is 10:00 UTC, a spread of 1.5× between the fullest and emptiest points of the day. The six busiest hours hold 29% of all observed broadcast time, against 25% for a platform spread perfectly evenly.

That is the finding, and it is not the one we expected: this is a genuinely round-the-clock market. A concentration of 29% is only 4 points off perfectly flat, and a 1.5× peak-to-trough spread is mild enough that there is no hour of the day that could reasonably be called off-peak. The industry's "always someone on" framing is usually read as marketing; measured against the most-watched rooms, it is close to literally true. The practical consequence for anyone choosing hours is that competition never really thins out — you cannot find a quiet slot to launch into, because there isn't one.

The audience clock

The same day, measured from the other side: mean concurrent viewers across the rooms we can see.

Mean concurrent viewers in the top rooms, by UTC hour · 204 hourly samples · PRELIMINARY, see limits below

Audience currently peaks at 22:00 UTC, but we are publishing that as an observation rather than a finding. It rests on 204 hourly samples, which is not yet enough for the per-hour means to have settled, and the honest thing to do with an unsettled number is label it. The comparison worth making — where audience is high relative to competition — needs both curves stable, and we will publish it when the second one is.

Turnover at the top

Because we log the first and last time we see each performer, we can measure something else nobody publishes: how quickly the top of the ranking changes hands. Note carefully what this is not — a performer leaving our data has not necessarily stopped broadcasting. She may simply have dropped out of the most-watched rooms. This is leaderboard turnover, not a quit rate, and anyone reporting it as the latter is misreading it.

Cohorts are excluded until enough time has passed to judge them, so "too recent" means unmeasurable, not zero. The 16,993 performers already broadcasting when we started are excluded entirely — they share an artificial first-seen date and would distort every rate on this table.

What the money looks like

Performers write their tip goals into their own room titles, which makes those figures the most honest price list in the industry — nobody is quoting them to a survey. Pooled across 118,495 figures parsed from stored room titles, the typical ask sits at 222 tokens, with the tenth percentile at 33 and the ninetieth at 1,770.

At the platform's published broadcaster rate of five cents a token, that median works out to roughly $11.10 to the performer per completed goal, against about $22.20 spent by the room. We walk through what that means as an income on the earnings page, and what tokens cost on the token calculator.

Method, and what would break it

Every figure above regenerates from the raw data, which is published in full and openly licensed. If you think we have this wrong, the fastest route to showing it is the download page.