Freelance Platform Fees vs Direct Clients: The Math

What is the total take rate once every fee is counted?

The total take rate is the commission plus payout fee plus FX spread, not the headline commission alone. On a low-fee marketplace, you may lose 12%-15% all-in. On a high-fee marketplace, especially one with a 20% seller charge, the stack can move into the low 20s before tax.

That gap is large enough to change your pricing. A freelancer who quotes from gross invoice instead of net cash usually underprices the job. If your budget assumes 100% of the invoice belongs to you, the spreadsheet is lying.

The table below uses working ranges from Upwork's pricing page, Fiverr's pricing page, Wise's pricing page, and Payoneer's fee schedule. I am not pretending the numbers are universal. They shift by corridor, method, and account type.

Here is a simple illustration for a $1,000 invoice paid through a marketplace and withdrawn into local currency. The exact figures depend on corridor, card rail, and withdrawal method, so treat the numbers as working ranges, not promises. Still, the order of magnitude is stable.

That is the version people skip. A 20% seller fee looks like the whole story until you cash out and convert. On a $1,000 invoice, another $15-$40 can disappear in FX, and the payout fee bites again. If you run the math on net cash, not gross invoice, the platform is more expensive than it first appears.

Do not compare the posted commission to your bank balance. Compare the invoice to the amount that lands in your operating account. That is the number that pays rent.

How much does FX spread add on top of platform commission?

FX spread usually adds 1%-4% on top of the platform commission, and it is easy to miss because the platform rarely labels it as a fee in the same way. If your invoice is in USD but you spend in UAH, the conversion step can matter as much as the platform cut. Wise's pricing page makes that spread more visible than many older payout systems.

The spread is the silent fee.

The clean way to measure it is simple. Take the mid-market rate at the time of invoice, compare it to the payout rate you actually receive, and call the difference by name. If the spread is 1.8% on paper but 2.6% on settlement day, use 2.6% in your model because that is the real cost of moving money.

Payoneer's fee schedule can also hide conversion cost inside the rate you receive. That does not make the fee fictional. It just means you need to compare the quoted payout amount to a mid-market reference and not to the platform's marketing copy. The same discipline applies to banks, card processors, and any payout tool that says conversion is included.

Small invoices bleed more. The percentage may be the same, but the fixed fee is a bigger bite when the job size is small. A $3 cash-out fee on a $150 invoice feels like friction; on a $5,000 invoice it feels like noise. The math changes with size, so the fee comparison has to change too.

If your platform commission is 10% and the conversion step takes another 2.5%, you are already at 12.5% before withdrawal charges. Add a fixed $3-$10 cash-out fee and the all-in cut on a $300 invoice looks much uglier than it does on a $10,000 one. That is why the same platform can feel acceptable for larger retainers and punitive for smaller jobs.

What does the platform actually provide for that fee?

The platform buys demand, escrow, and a default dispute path. That is the real product. The fee is not only a toll on your work; it is also payment plumbing, buyer trust, and a small amount of enforcement that you would otherwise have to build yourself.

Upwork's pricing page and Fiverr's pricing page are useful precisely because they expose different bundles of that bundle. One platform may charge less on the front end but give you less buyer access. Another may charge more and still be worth it if it keeps the invoice moving. The comparison only works if you name the services being purchased.

Archive depth is mostly dead weight.

A pile of old reviews helps, but it matters less than this week's buyer flow. What pays rent is current discovery, not a museum of past wins. If the marketplace still puts active buyers in front of you and shortens the sale cycle, the fee is buying time as well as access.

• Discovery: buyers already search there.
• Escrow: money is held before work starts.
• Dispute handling: there is a default process when scope breaks.
• Payout routing: you do not wire money manually after every job.

That bundle is why some freelancers keep a marketplace account even after they build direct sales. The platform is not a religion. It is a distribution rail with rules. If you already have a warm referral channel, the fee is easier to question. If you do not, the platform is doing real work.

For first-time buyers, the platform can also reduce the time you spend explaining why a deposit is normal. That is not glamorous, but it saves hours. If a client already knows the checkout flow, you spend less energy on trust and more on delivery.

At what income level does going direct become worth the risk?

Going direct becomes worth the risk when your direct acquisition and collection cost stays below the all-in platform stack by at least 5 points and one unpaid invoice will not wreck your month. For a solo freelancer, that often means the direct path is attractive once 30%-50% of monthly revenue can come from repeat buyers outside the marketplace.

The numbers move, so treat that as a rule of thumb. If the platform path costs 18%-25% all-in and your direct path costs 8%-12% in sales time, invoicing, payment fees, and bad-debt reserve, direct starts to win. If your direct pipeline is weak, the platform is still cheaper than a cold month.

Break-even is not a slogan. It is a formula: platform fee plus payout fee plus FX spread minus direct acquisition cost minus expected collection loss. If direct costs 10% and the marketplace costs 22%, you have a 12-point spread to spend on outreach, follow-up, and the invoice work you now own. If that spread disappears, the marketplace wins again.

A 10% platform fee can be cheaper than a 0% direct deal. If the marketplace brings escrow, a buyer who already trusts the checkout flow, and a clean dispute path, you may save more than 10% of your time and avoid the one invoice that never clears.

That sounds backwards until you price bad debt. One unpaid $2,000 invoice wipes out the savings from many smaller direct wins. If your direct close rate is thin or your client quality is noisy, the platform fee is insurance you are already using whether you name it that way or not. A single lost invoice can also freeze your cash flow for weeks, which is a cost the headline fee never shows.

The cleanest trigger is not revenue alone. It is repeatability. Once you can replace a missing marketplace order without panic, you can start shifting work off-platform in controlled steps. Until then, the direct move is usually bravado with nicer branding.

How do direct clients pay, and what rails do they accept?

Direct clients usually pay by ACH, bank wire, SEPA, Wise, PayPal, or Stripe invoice. The rail they accept depends on geography, company size, and who sits in accounts payable. Ask for the method that their finance team can clear fastest, not the one you like best.

Cash flow is a rail choice.

• ACH: usually the cheapest for US buyers and often the cleanest if both sides are in the United States.
• Bank wire or SWIFT: common for larger B2B invoices and cross-border work, though the bank fees can be blunt.
• Wise: often a good middle path for cross-border invoices when both sides want visible conversion cost, per Wise's pricing page.
• PayPal or card via Stripe: familiar to small buyers, but usually pricier once fees and FX stack up.
• SEPA: the normal answer for euro clients, especially when the buyer is in the EU.

Mid-market buyers usually accept bank wire because accounts payable already has a playbook. Smaller buyers often want card or PayPal because they are trying to spend from a card on file. Your job is to map the rail to their process, then keep the invoice simple enough that nobody opens a support ticket to pay you.

For first-time direct clients, ask for 30%-50% up front. For repeat clients, net 7 or net 15 is usually enough if your work is specific and the buyer is real. Net 30 is a financing decision, not a neutral term, and you should price it like one. If they ask for net 60, you are lending them money.

Payoneer's fee schedule and Wise's pricing page are the two references I would check first if I were comparing cross-border payout paths. Both can be cheaper than a careless wire chain, but neither is free. The winner is the rail that gets money from their bank into yours with the fewest hidden steps. If you can hold funds in the invoice currency, you can reduce forced conversion, but that shifts FX risk onto you.

What do you give up by leaving the platform's protection?

You give up escrow, built-in dispute handling, and the platform's reputation layer. You also give up the default screening that keeps some bad buyers out and the easy replacement flow when a client disappears. That is the real cost of going direct.

Protection is a service.

What replaces it is manual discipline. You need a written scope, deposit terms, milestone dates, acceptance criteria, late-fee language, and a stop-work rule when scope changes. DIY monitoring works, and almost nobody sustains it, but that does not make it optional. The direct system fails when the freelancer wants to act like the platform still exists.

The best direct-client setup is boring. It uses a contract, a deposit, an invoice, and a follow-up schedule. It does not rely on goodwill, because goodwill is not a collection method. It also uses one source of truth for the brief, so scope changes cannot hide in email threads.

You also lose some speed. On a marketplace, the buyer already knows the checkout path and the platform already knows how to move money. Direct work asks the client to trust your process instead, which is why a direct offer usually converts better after you have already completed one project on-platform or through a warm referral. The second sale is cheaper than the first.

You still can build this by hand. It just takes more follow-up than most freelancers expect, and that follow-up is work, not atmosphere.

Bottom line: if the platform stack is eating 15%-25% and you cannot replace that demand cleanly, stay where the work is. If you can win repeat buyers, control collection, and keep payment friction below the marketplace toll, direct becomes the better trade.

Frequently asked questions

Is a 20% marketplace fee always worse than direct?

No. A 20% fee can still be cheaper than a bad direct deal. If the platform gives you escrow, buyer trust, and a payment path that works, you may save more than 20% of your time and avoid the invoice that never clears.

What fee should I model first?

Model the payout route first. Commission is visible, but FX spread and withdrawal fees are where spreadsheets break. If you invoice in USD and spend in UAH or another local currency, the conversion step can move your net more than the headline fee.

Which payment rail is cheapest for direct clients?

ACH is usually the cheapest rail in the US. For cross-border work, Wise, SEPA, or a bank wire may be cleaner depending on corridor and invoice size, while PayPal and card rails are easier to sell but usually cost more.

When should I start pitching direct?

Start when one missing invoice will not break your month. That usually means repeat buyers, a cash buffer, and a direct pipeline that can replace 30%-50% of marketplace revenue. Until then, keep direct as a parallel track, not a forced migration.