Researchers Interviewed 27 OnlyFans Creators. The Constraint Wasn't Content.

A three-year study of how creators actually get found suggests the work that decides who earns happens entirely off the platform.

Of the 27 OnlyFans creators interviewed for a study published last summer, exactly one had a direct link from a mainstream social media account to their paid page. The other 26 routed followers through intermediaries — link aggregators, Telegram channels, backup profiles — because posting the link openly tends to get the account removed.

That single finding explains a lot about a platform most people misread.

The study, published in the SAGE journal Platforms & Society by Margherita Di Cicco and Davide Beraldo, combined digital ethnography with in-depth interviews conducted between September 2021 and December 2023. Nineteen participants identified as women, eight as men; they were based in Italy, France, Denmark, the UK and the US, and the interviews ran an average of 78 minutes. The research was approved by the doctoral committee at the University of Milan.

Its central question was not how much creators earn. It was what they spend their time doing. The answer, across every participant, was marketing — and the researchers conclude that this promotional labour matters more to a creator's outcome than the content itself.

The design decision underneath everything

OnlyFans does not let you browse. There is no category view, no recommendation feed, no way to stumble across a profile. Access is by direct link, which means a creator's audience has to be assembled somewhere else and then walked across.

Di Cicco and Beraldo give this a name: "cross-platform dependency by design." The phrase does two jobs. It identifies the dependency — creators cannot rely on the platform that pays them for the audience that pays them — and it locates the cause in architecture rather than accident. This is not a feature the company forgot to build. It is the shape of the product.

The consequence is that success on the platform is decided before anyone arrives on it.

The maze, mapped

The researchers reconstructed each participant's network of platforms and the links connecting them, then visualised the result. What emerges is not a funnel. It is a warren.

Most creators send followers from Instagram or TikTok to a link aggregator, and from there onward. Many insert Telegram as a buffer step, and Telegram turns out to occupy a distinctive position — in the study's mapping it was the one platform participants would link to their paid page directly. Some maintain backup accounts whose only function is to hold the link. Many paths are deliberately redundant, with the same destination reachable from several directions, so that losing one account does not sever the route.

Several of the maps are circular: social media leads to the paid page, and the paid page leads back to social media.

For a reader trying to find a specific creator, this architecture is the whole explanation for why it feels so difficult. The path is not obscure by accident. It was built deliberately indirect by people managing the risk of being deleted.

Why a 2018 law is in this story

The hostility the study documents has a legal origin. FOSTA-SESTA, signed into US law on 11 April 2018, carved sex-trafficking enforcement out of the Section 230 protections that shield platforms from liability for what users post.

Platforms responded the way anyone facing new liability responds: by over-correcting. Community standards tightened around nudity and sexual expression specifically, and the enforcement that followed has been consistently described by researchers as opaque, inconsistent and applied unevenly.

The participants in the study describe the practical version of this. Rules that turn on whether swimwear appears near water. Accounts that vanish and have to be rebuilt from zero. Captions written to mislead automated moderation, with no confidence that they work. One participant recounted losing and recreating well over a dozen profiles across a single year before being unable to register new ones at all.

None of this is content moderation as most users experience it. It is a working condition.

The part that answers the earnings question

Buried in the moderation findings is the mechanism that converts fame into money.

Participants reported - and the researchers connect this to earlier work by Carolina Are and Susanna Paasonen — that prominent accounts are policed more leniently than everyone else. Comparable material from a celebrity and from a sex worker does not attract comparable enforcement. Where the celebrity's post survives, the sex worker's is suppressed.

Stack that against the architecture and the picture completes itself. Visibility must be built off-platform. Off-platform visibility is governed by systems that suppress this category of creator. Those systems suppress it less if you are already famous. So the people who arrive with an audience keep theirs, and the people who arrive without one are trying to build it in the narrow space the moderation leaves open.

That is a structural answer to a question usually given a behavioural one. The vast content industry offering advice on succeeding as a creator - post more often, reply faster, pick a niche - is describing the small controllable residue of a process largely decided by infrastructure.

There is a further irony the earlier literature captured. A 2023 study of 22 new sexual content creators presented at the ACM CHI conference found that the platform's mainstream visibility, and the participation of well-known figures, was what made joining feel socially acceptable in the first place. Visibility drew people in. Visibility is then the thing they cannot get.

The gap this leaves has been filled from outside, by parties with no stake in the platform's design. Independent indexes such as spicycreator.com compile public profile information into the category and location filters that do not exist natively, which is a partial substitute for a browse function and, for anyone who does not already have a name in hand, the only one available.

What the study is careful about

It is qualitative work, and the authors do not pretend otherwise. Twenty-seven participants recruited by snowballing, mostly in their twenties, mostly European — that is a sample designed for depth, not for projection onto four million accounts. The coding was done by one researcher, which the paper acknowledges directly.

What it offers is not a statistic. It is an account of the mechanism, from the people inside it, which is precisely the thing that platform-level data cannot show.

The authors state their aim plainly enough: to push back on the description of this work as easy money. On the evidence they gathered, most of the work is not even the work most people assume it is. It is unpaid marketing, performed on platforms that would rather it stopped.