Ask a freelancer what Upwork costs and you will usually get a single percentage. The single percentage is the smallest part of the answer. Between application costs, the service fee, payment processing, withdrawal charges and currency conversion, the gap between your quoted rate and the money in your bank account is consistently wider than people assume.
This is not an argument that Upwork is a bad deal. It is an argument for knowing the real number, because you cannot price sensibly against a figure you have never calculated.
Where money leaves along the way
There are five separate points at which your earnings shrink. Only the second one gets discussed.
- Connects. You buy tokens to submit proposals. Competitive listings cost more connects, and most proposals do not convert — so this is a real, ongoing customer-acquisition cost, paid whether you win or not.
- Service fee. A percentage deducted from every payment a client makes to you. Upwork has restructured this more than once; check the current rate on your own contracts rather than trusting any article, including this one.
- Client-side fees. Not deducted from you, but paid by your buyer on top of your rate. It comes out of the same budget, so it depresses what clients are willing to agree to.
- Withdrawal fees. A per-transfer charge that varies by method. Small on a large monthly withdrawal, painful if you withdraw weekly.
- Currency conversion. The quietest cost of all. If you earn in USD and bank in another currency, the spread applied on conversion is frequently larger than the withdrawal fee, and it never appears as a line item.
Work out your own effective rate
Rather than reciting percentages that will be out of date by the time you read this, run the calculation on your own account. Take one real month and fill in this table from your transaction history:
| Line | Where to find it | Your number |
|---|---|---|
| A. Gross contract earnings | Reports → earnings for the month | — |
| B. Service fees deducted | Transaction history, fee rows | — |
| C. Connects purchased | Membership & connects billing | — |
| D. Withdrawal fees | Transaction history, withdrawal rows | — |
| E. FX loss | Amount withdrawn × (mid-market rate − rate received) | — |
| F. Hours worked including proposals | Your own tracking — be honest | — |
| Effective rate | (A − B − C − D − E) ÷ F | — |
Line F is the one people leave out and the one that decides everything. If you spend six hours a week writing proposals and win one in twelve, that time is part of the cost of every job you do win. A freelancer billing $60/hour with a heavy proposal habit can easily be earning $38/hour in reality.
Compare the result against the floor rate you calculated when setting your prices. If the effective rate is below your floor, the platform is not a channel — it is a subsidy you are paying.
Making the fees smaller
Several of these costs are partly within your control.
- Favour long contracts over many short ones. Fee structures on marketplaces have historically rewarded ongoing relationships, and one long engagement carries a fraction of the proposal cost of ten small ones regardless.
- Apply less, better. Ten researched proposals beat sixty templated ones on both conversion and connect spend. Read the whole listing; if the client’s budget is a third of your rate, that is information, not a challenge.
- Withdraw monthly, not weekly. Fixed per-transfer fees are pure arithmetic.
- Fix your currency route. A multi-currency account that receives USD directly, rather than auto-converting at whatever rate your bank chooses, often recovers more money than any negotiation you will have this year.
- Price the fee in. Your platform rate should not be your direct-client rate. If the platform takes a percentage, your listed rate needs to be higher to net the same amount — this is normal, and every experienced freelancer does it.
A worked example
Numbers make the point better than percentages. Take a freelancer with a listed rate of $60/hour who bills 90 hours across a month on marketplace contracts, and who spends six hours a week writing proposals.
| Line | Amount |
|---|---|
| Gross billed (90 h × $60) | $5,400 |
| Platform service fee | −$540 |
| Connects spent on proposals | −$45 |
| Withdrawal fees (2 transfers) | −$14 |
| Currency conversion spread (~1.5%) | −$72 |
| Net received | $4,729 |
| Hours billed | 90 |
| Hours on proposals and admin | 26 + 12 |
| Effective rate (128 hours) | ≈ $37/hour |
A listed rate of $60 becomes an effective rate of $37. Nothing here is hidden or dishonest; every line is disclosed somewhere. But almost nobody adds them together, and the person quoting $60 usually believes they are earning close to $60.
Notice which line does the most damage. It is not the service fee — it is the 38 unbilled hours. Improving your proposal-to-win ratio moves the effective rate far more than any fee optimisation available to you.
What the fee legitimately buys
It is worth being fair about the other side of the ledger, because “platforms are a rip-off” leads people to abandon a channel that is still profitable for them.
- Buyers you could not reach. Companies searching a marketplace are not going to find your website.
- Escrow and payment protection. Genuinely valuable when working with strangers across borders, and hard to replicate yourself.
- Dispute processes and identity verification. Imperfect, but better than nothing when a client vanishes.
- Invoicing, contracts and compliance paperwork handled for you across jurisdictions.
- A public track record that new clients can verify, which is enormously valuable when you have nothing else.
Priced against the cost of acquiring an equivalent client yourself — the hours of outreach, the unpaid pitching, the risk of non-payment — a commission on early work is often a bargain. It is later, when clients would find you anyway, that the same fee becomes pure loss.
The rules you should not break
The obvious way to avoid fees is to take a client off-platform. Do not do this while the contract is live and covered by the platform’s terms. The consequences are severe and one-directional: account suspension, loss of your entire review history, and forfeiture of funds in escrow. Your ratings and history are the only asset you have accumulated there, and they are non-portable.
There is usually a legitimate path — Upwork has historically offered a paid buyout that converts a marketplace relationship into a direct one. If a client relationship is genuinely long-term, that fee is often recovered within a couple of months. Read the current terms and use the sanctioned route.
The fee is not the problem. Being unable to leave is the problem.
When the platform stops paying for itself
A marketplace fee buys you distribution — buyers you could not otherwise reach, plus escrow protection that genuinely has value when you do not know the client. Early on, that is an excellent trade. You are converting money into reputation, and reputation is the scarce resource.
The trade turns bad at a specific, identifiable moment: when you are paying commission on clients who would have found you anyway. Watch for these signals:
- Most of your platform income comes from two or three repeat clients
- You are declining work because you are full
- People approach you by name rather than through a listing
- Your effective rate has been flat or falling for two quarters
Three of those four and the commission has stopped buying anything. The right response is not to quit abruptly — it is to start building an owned channel now, while the platform income is still paying your rent. Six channels that do not involve bidding is the practical starting point, and the platform comparison covers whether a different marketplace would suit you better in the meantime.
Run the effective-rate calculation once a quarter. It takes twenty minutes and it is the only honest way to know whether the arrangement is still working for you or merely familiar.
Leave a Reply