Most complaints about Upwork start with the percentage it takes from your earnings. The cost that arrives sooner gets far less attention: you pay to apply. Upwork sells Connects at $0.15 each, and every proposal burns some of them. A single proposal usually needs somewhere between 2 and 16 Connects, depending on the job, which puts one application between $0.30 and $2.40. You pay it whether the client reads your pitch, ignores it or hires somebody else.
That sounds trivial until you send forty proposals in a month. This guide works through what those applications really cost, including the part that never shows up on a receipt, which is your own time. Every figure below is an illustration built from stated assumptions, so swap in your own.
From cost per proposal to cost per win
What matters is not what one proposal costs. It is what one won contract costs, because the Connects you spent on every losing proposal have to be covered by the ones that win. The sum is short: divide the cost of a proposal by your win rate.
| Win rate | 6 Connects ($0.90) | 12 Connects ($1.80) |
|---|---|---|
| 5% | $18.00 | $36.00 |
| 10% | $9.00 | $18.00 |
| 20% | $4.50 | $9.00 |
The win rates here are assumptions, not Upwork statistics. The pattern holds regardless. Double your win rate and every contract you land costs half as much to get. Apply to jobs that swallow twelve Connects instead of six and the price doubles.
The cost nobody invoices
Now add the minutes. A proposal worth sending takes real work: reading the post twice, checking the client's history, writing something that could only have been written for this job. Call it twenty minutes. If an hour of your time is worth $40, those minutes cost $13.33. Add six Connects ($0.90) and one proposal costs $14.23. At a 10% win rate, each contract you win has cost about $142 to land.
Look at where the money went. The Connects are $0.90 of the $14.23, about six percent. Almost everything else is time. That changes the advice. Hoarding Connects barely helps; it is time that drains the account.
When extra effort pays and when it doesn't
You might assume a longer, better proposal always wins. It only pays if it lifts your win rate roughly in proportion to the time you add. The table compares five styles of working, all with six Connects and a $40 hour.
| Time per proposal | Win rate | Cost per won contract |
|---|---|---|
| 10 min | 5% | $151.33 |
| 20 min | 10% | $142.33 |
| 30 min | 15% | $139.33 |
| 30 min | 20% | $104.50 |
| 45 min | 20% | $154.50 |
Ten-minute proposals at a 5% win rate cost about as much per win as thirty-minute ones at 15%. Thirty minutes at 20% is clearly better, and forty-five minutes at 20% is clearly worse. The lesson isn't "write more" or "write less". Measure your own win rate against the time you spend, then keep the style that gives the lowest number.
Contract size decides whether any of this works
Take the baseline cost of $142.33 per win and set it against what you earn from the contract.
| Contract value | Cost to land it as a share of the contract |
|---|---|
| $100 | 142.3% |
| $500 | 28.5% |
| $2,000 | 7.1% |
| $5,000 | 2.8% |
On those assumptions a $100 job loses money before you write a line of work, and a $500 job hands over more than a quarter of its value to the process of winning it. Past $2,000 the cost shrinks to something you can absorb. That's why many experienced freelancers set a minimum job value and skip the rest, even when they could technically do the work.
Ways to bring the number down
- Apply less often, to better matches. If you meet fewer than most of a job's requirements, the odds are poor and the minutes are gone.
- Check the Connects price before you start. A job asking for sixteen deserves more scrutiny than one asking for four.
- Build pieces, not templates. A reusable opening that clients recognise as a template tends to lose. Reusable proof (a relevant example, a short process note) saves time without sounding canned.
- Read the client, not just the post. History of hires, payment verification and how they write all tell you if the contract will actually pay.
- Raise your floor. Skipping the small jobs improves the maths faster than any writing trick.
A worked month
To see how the pieces fit, imagine a month with 60 proposals at six Connects each, a 10% win rate and twenty minutes per proposal. You spend $54 on Connects and twenty hours writing. At $40 an hour that's $800 of time, so $854 in total to land six contracts. If each is worth $500, you've earned $3,000 and spent 28.5% of it just winning the work. If each is worth $2,000, the same effort brings $12,000 and the cost falls to 7.1%. Same effort, wildly different return, which is why the size of the contracts you chase matters more than any single proposal tweak.
A tracking sheet worth keeping
Keep a simple spreadsheet for a month: the date, job type, Connects used, minutes spent, outcome and contract value. At the end, add up Connects, add up the value of your time, and divide by the number of contracts won. If you landed three contracts and spent $9 in Connects and eleven hours writing, you know your true cost of winning work, and you can compare job types against each other. Some categories will clearly be worse than others, and dropping them is free money.
When Connects stop being worth it
If your typical contract is small, your win rate is stuck low, and you see no improvement after thirty proposals, the platform is costing more than it returns. That isn't a failure. It's information. Many freelancers move to direct clients or niche communities once they have a few reviews, where the terms of each platform allow it.
To turn all of this into take-home pay, run a contract through the net earnings calculator, then see what remains after tax in the take-home pay calculator. The Upwork fees guide covers the service fee itself.
Last reviewed October 2026. Figures are estimates for planning, not tax advice. See the disclaimer and methodology.