Ask what a chatter costs and the answer that comes back is an hourly rate. The hourly rate is a real number and it is not the cost. An agency does not buy a person, it buys the property that a specific inbox is answered at every hour of every day, and that property is expensive in a way an hourly rate conceals, because it is bought in whole humans working whole shifts.
The arithmetic of a covered inbox
Start from the shift structure the reporting describes. The BBC's account is eight hour shifts, five days a week.1 A week has 168 hours. Covering all of them in eight hour shifts takes 21 shifts, and a worker on a five day week supplies five of them. So one continuously covered seat needs 4.2 full time workers, before anyone takes a holiday, is ill, or leaves.
| Published hourly rate | Source of the rate | 168 hours (one week) | 728 hours (one month) |
|---|---|---|---|
| $2.00 | BBC News, 2026, first agency | $336 | $1,456 |
| $3.00 | Fortune, 2024 | $504 | $2,184 |
| $4.00 | BBC News, 2026, second agency | $672 | $2,912 |
Hourly rates: BBC News, 11 March 2026 (source 1) and Fortune, 5 April 2024 (source 2). Totals are 168 and 728 hours multiplied by the published rate, computed by us. They are wages only. They exclude every item in the next section.
Two things are visible immediately. The first is that even at the lowest rate any journalist has documented, continuous coverage of a single inbox is a four figure monthly line. The second is that the line does not move when revenue moves. A quiet month costs the same as a good one, which is the defining property of a coverage cost and the reason chatting teams are sized to the best week rather than the average one.
Wages scale with the clock. Revenue scales with the readers. The two do not move together, and the gap is where agencies fail.
The costs that are not the wage
None of the following is published as a figure anywhere we could find, so none of it is given a number here. All of it is visible in the structure the reporting describes.
Supervision
Chatters are managed. Vice describes account managers overseeing around 10 to 15 creators each and taking 15 to 25 percent off of each creator.3 That is a documented commission layer sitting on top of the wage layer, and in the accounts where it appears it is a percentage of creator revenue rather than a fixed cost, which means it grows precisely when the operation succeeds.
Load, and what load does to quality
Cost per hour falls when a worker holds more threads, and quality falls with it. Rest of World describes workers chatting with "a half dozen fans simultaneously", in an operation covering up to 50 creators, under quotas that had risen 25% in six months.4 Vice describes one worker on as many as seven creators at any one time.3
The savings from increasing that ratio are immediate and legible on a spreadsheet. The cost shows up as the failures described in objections: forgotten context, repeated questions, threads that reset. That cost is real and lands on a later month, which is why the ratio tends to rise until something visibly breaks.
Turnover and training
No public source publishes a turnover rate for this role, so this site does not claim one. What is documented is the combination that produces turnover everywhere else it appears: low pay against a local benchmark,4 night hours, quotas that rise,4 and work a union has described as involving exposure to "potentially egregious or harmful content".1 Each replacement carries a hiring cost, a training cost, and a period during which the new worker is running the sequence worse than the person who left.
The hand-over
Every shift boundary costs something: the notes written, the notes read, and the context that did not fit in either. This is pure overhead created by the fact that coverage is assembled from humans, and it is one of the few costs that grows linearly with the number of shifts rather than with revenue. It is described in more detail in timing and time zones.
What the software costs, for comparison
The other side of this comparison is public, because the tools publish prices. These are reproduced exactly as their pages stated them on the date shown, with no commentary on value.
| Tool | Published price | Page read |
|---|---|---|
| OnlyChat | Pro $20 per creator per month plus 10% commission; Premium $70 per month plus 10%; Telegram $20 per month plus 15% commission | 2026-09-14 |
| Other published tools | Typically a monthly fee per creator or per bot, from $0 to about $99, plus a commission of roughly 5% to 20% on AI sales; some publish no price at all | 2026-09-14 |
The OnlyChat row is source 5 below, as printed on its pricing page on 14 September 2026. The second row summarises, without naming them, the pricing pages of other tools read the same day. Prices change; check the page before relying on any of them. A short selection is on tools chatters use; the full ranking is at chatbot-ofm.com.
One kind of number found on vendor sites in this market is a claim rather than a price, and belongs in a different category: a percentage by which the product is said to be cheaper than humans, published by the vendor, with no method attached. That is a vendor claim about a comparison, and it is the shape of claim a reader will meet everywhere in this market. It is not evidence of anything, and this site does not reproduce such figures or repeat them as findings.
The honest summary
What can be said from published sources: continuous coverage of one inbox requires roughly four full time people on the shift patterns journalists describe; documented hourly rates for the role run between under $2 and about $4; a supervision layer of 15 to 25 percent of creator revenue sits above the wage layer in at least one documented agency; and the ratio of threads per worker is rising while quotas rise with it.
What cannot be said from published sources, and is therefore not said here: turnover rates, training costs, the revenue difference between a good chatter and an average one, and any figure comparing the total cost of a human team with the total cost of software. Those comparisons are made constantly in this industry and they are made with numbers nobody has published.
The direction the industry has taken, and the facts on both sides of it, are in human chatting or AI chatting.
Sources
- BBC News. Chris Vallance, “Icky and heartbreaking”: The $2 per hour worker behind the OnlyFans boom. https://www.bbc.co.uk/news/articles/cq571g9gd4lo published 2026-03-11, read 2026-09-14
- Fortune. Alexandra Sternlicht, AI is coming for the OnlyFans chat industry. https://fortune.com/2024/04/05/ai-is-coming-for-the-onlyfans-chat-industry published 2024-04-05, read 2026-09-14
- Vice. Eloise Hendy, Think You're Messaging an OnlyFans Star? You're Talking to These Guys. https://www.vice.com/en/article/onlyfans-management-agency-chatters/ published 2023-10-24, read 2026-09-14
- Rest of World. Michael Beltran, A hidden network handles chats for OnlyFans stars. AI could soon take over. https://restofworld.org/2025/onlyfans-ai-dm-bots/ published 2025-08-20, read 2026-09-14
- OnlyChat. OnlyChat pricing. https://only-chat.ai/pricing read 2026-09-14