Every marketplace is trying to solve the same basic problem. Two strangers want to do business, and neither side has a good reason to go first.
The buyer will not pay before the work exists. The seller will not work before the money exists. If nobody steps in, the transaction can easily stop before it starts.
The marketplace steps into that gap. It becomes the party both sides trust instead of each other. It holds the money, sets the rules, decides what happens when things go wrong, and carries the reputation history that makes a stranger possible to assess.
That is a real service, and it deserves to be paid for. The useful question is whether the way marketplaces charge for it has much connection to the cost of providing it.
A marketplace fee pays for a few different jobs
Break a typical freelance marketplace commission into the work it funds, and you get roughly five categories.
- Escrow, meaning the marketplace holds funds between agreement and delivery. The cost is payment processing plus the engineering needed to do it correctly.
- Identity and vetting, meaning the marketplace establishes that people are who they claim. The cost is real, and mostly one-off per account.
- Dispute resolution, meaning humans decide what happens when parties disagree. This is expensive and valuable.
- Discovery, meaning the marketplace helps buyers find sellers. It is expensive to build, cheap to run, and the cost does not rise with transaction size.
- Reputation, meaning accumulated history that makes strangers legible. It is nearly free to store and the most valuable thing on the list.
Now look at how the fee is usually charged. It is a percentage of transaction value, typically between five and twenty percent, on every job, forever.
The price grows faster than the cost
Almost none of those costs scale with transaction value.
Escrow on a five thousand pound project costs a little more than escrow on a five hundred pound one, because payment fees are partly proportional. Everything else is flat, or close to flat.
Verifying an identity costs the same regardless of what the person later earns. Storing a review costs the same. Search costs the same.
Dispute resolution is the real exception, because people may need to spend time deciding a case. Even there, disputes are rare. On the platforms that publish figures, they are low single-digit percentages of transactions.
So a percentage fee is best understood as a tax on the value of the work. It is collected by the party that happens to sit between the two people creating that value.
The mismatch gets worse in long relationships
The mismatch compounds in a specific and unpleasant way.
Consider a freelancer and a client who work together for three years. The marketplace introduced them once. That introduction provided real value.
After that, the marketplace continues to take a percentage of every transaction for the next three years. During that period, it is providing escrow and very little else.
Both parties know this. That is why every marketplace of this type has a large, invisible economy of people who met on the platform and moved off it.
It is also why every such platform has aggressive terms prohibiting exactly that. The prohibition is an admission: the fee has drifted far enough from the value that both sides are motivated to escape it, and only a contract keeps them in.
A pricing model that requires a legal clause to stop its own customers routing around it is telling you something.
The strongest defence is acquisition cost
The strongest defence of percentage pricing is acquisition, rather than cost recovery.
Marketplaces are enormously expensive to start, because they are useless until both sides are present. Getting through that is a years-long, capital-intensive effort.
The percentage fee on successful transactions is what eventually pays for that effort. Charging small flat fees from day one would never fund it.
This is true, and it is the honest reason the model exists. But it describes a way of financing a business, rather than a price for a service. Those two things get conflated deliberately.
Once the network is built and the acquisition cost is sunk, the percentage keeps running because it can.
There is a second argument: percentage fees align incentives, because the platform earns more when its users earn more. There is something to this. It is also why platforms optimise for transaction value rather than for the smaller, faster jobs many users would prefer.
A zero commission only works if the functions are still paid for
If you take the commission away, the functions still have to be paid for. This is where most zero-fee ideas collapse, so the conditions matter.
Four things have to be true.
- Escrow has to cost almost nothing to run. That means holding funds in a way that is correct, auditable and cheap: proper state modelling, automated release conditions, and payment rails chosen for cost rather than convenience. It takes engineering effort up front, with near-zero marginal cost afterwards.
- Disputes have to be rare by design. Cheap adjudication is weaker than designing the disagreement out. The moment of agreement carries the weight: clear scope, defined acceptance, and staged milestones so any disagreement is small and early. Every dispute you design out is one you never pay to resolve.
- Revenue has to come from something optional. The money comes from services people choose to buy on top: verification, promotion, tooling, larger organisational features. If the core is free, the money has to be genuinely additional rather than a toll wearing a different name.
- Reputation has to be portable. This is counter-intuitive, but essential to the argument. If the platform’s hold on users is that they cannot take their history elsewhere, it is competing on lock-in and the fee will creep back. Portability forces it to compete on being useful.
The freelancer’s arithmetic is hard to ignore
Percentages are easy to wave through in the abstract, so put a person in it.
A developer bills sixty thousand a year through a marketplace charging ten percent. That is six thousand pounds.
That is not a one-off introduction fee. It is an annual charge, repeating for as long as they use the platform.
Over five years, that is thirty thousand pounds. The service’s marginal cost to provide, once the platform exists, is dominated by payment processing at well under one percent. The gap is a multiple.
Now consider what the same person would happily pay for services they actually value.
- Verified identity that shortens their sales cycle.
- A portfolio that ranks well.
- Tooling that handles their contracts and invoicing.
- Priority placement when they want it.
Several hundred a year, willingly, for things that help them earn. People object because they are charged for the wrong thing, in the wrong shape, forever.
The fee changes how people behave
Prices change conduct, and a percentage fee changes it in ways that hurt the marketplace itself.
- It pushes work off-platform. The larger and longer the engagement, the greater the incentive to leave. The platform systematically loses its best relationships and keeps its most transactional ones, which is precisely backwards.
- It distorts pricing. Freelancers inflate quotes to absorb the fee, so buyers pay more than the market rate and blame the freelancer rather than the intermediary.
- It penalises small jobs. A twenty percent fee on a two hundred pound job is the difference between worth doing and not. The long tail of small, fast work that would make a marketplace useful never forms.
- It creates an adversarial relationship with the platform’s own rules. Users spend energy working out what they can get away with. Platforms spend energy on enforcement. None of that effort produces anything for anyone.
The replacement revenue has to be optional
Removing the commission is only credible with an answer to what replaces it. Ours is four things, all optional, none charged per transaction.
- Verification. Identity, right to work, insurance and qualification checks, paid by the freelancer who wants to shorten their sales cycle or by the client who wants the assurance. Priced at what the checks cost plus a margin.
- Placement. Paid visibility, clearly labelled. This is the difficult one, because it has to be done without corrupting search quality, and the discipline is to keep the paid slots few and obvious.
- Tooling. Contracts, invoicing, time tracking and tax summaries. Boring, useful, and priced as a subscription that people can cancel without losing access to the marketplace itself.
- Teams and organisations. Companies managing many freelancers need approval flows, consolidated billing and reporting. That is real software with real value and it is where the largest single revenue line sits.
Every one of those is something a user can decline while still transacting. That constraint keeps the pricing honest, and it is why we publish the revenue model rather than leaving people to guess.
We built Open Lance around that model
Open Lance exists because we thought those four conditions were achievable and nobody was trying. We had also built enough payment and escrow systems for clients to have a view on the engineering.
The commission is zero. Freelancers keep what they earn. Escrow, identity and dispute handling are funded by optional services, rather than by a slice of every job.
That is a harder business to build than a percentage. It requires the escrow to be cheap to operate, rather than merely profitable at fifteen percent. It also removes the comfortable revenue that arrives whether or not you are still adding value. Both of those are the point.
We misjudged two parts of the marketplace problem
Two things are worth saying because the argument is stronger with them in it.
- We underestimated how much of a marketplace’s value is discovery rather than trust. Trust is what we engineered. Finding the right person is what users actually struggle with, and a zero fee does nothing for that. It is the bigger problem and we were slow to admit it.
- We also underestimated how suspicious people are of free. A zero commission reads to many freelancers as either a temporary promotion or evidence that the money is being taken somewhere they cannot see. The honest response is to publish exactly where revenue comes from, which we now do, and to accept that trust in a trust machine takes years rather than a launch announcement.
The same pattern appears beyond freelancing
This is about more than freelancing. Any intermediary that charges proportionally for services that cost flat is running the same arrangement.
There are a lot of them.
- Payment processors above a certain volume.
- App stores.
- Booking platforms.
- Agencies taking a percentage of media spend.
In each case the question is the same. Is the fee a price for something, or a position rent collected from the value other people create? The test is simple: if both parties would leave the moment they were allowed to, the answer is the second one.
We think that arrangement is under more pressure than it looks, and that the platforms which move first to pricing that reflects cost will hold their users without needing a clause to do it. If you want the full argument with the economics attached, the Open Lance case study sets out what it took to build.
