Freelance rates: the number that matters isn't the one you quote

Most rate advice treats the rate as the finish line. Raise it, defend it, stop apologizing for it. The largest European freelance rate study that publishes its method puts the average hourly rate at €103 in 2026, down slightly from €104 the year before and up from €96 in 2022. Read the method and the number narrows fast: about 5,400 respondents, heavily weighted to IT and engineering contractors, all in Germany, Austria and Switzerland. It is a real number about a specific population. It still tells you nothing about what a single one of those contractors kept.

That is the flaw in treating the quoted rate as a measure of anything. It is the price of one input, agreed before the project had a chance to disagree with you.

A quoted rate is a hypothesis, not a result

When you say €80 an hour, you are making a prediction: that the hours you bill will be roughly the hours the work consumes. Every hour the project takes that never appears on an invoice is a hole in that prediction, and the money leaks out of it at exactly the rate you quoted.

The number on the other side is your effective hourly rate: the fee actually collected, divided by every hour the engagement consumed from first email to cleared payment. Not billable hours. All hours. Scoping calls, the proposal you wrote on a Sunday, the third revision round nobody priced, the week you held capacity while the client's legal team read the contract, the invoice you sent twice.

Freelancers who quote well and measure badly can go years without noticing the difference, because both numbers feel like income. One of them is.

Where the other 53 hours went

Take a €6,000 fixed-fee project, quoted at €80 an hour against an estimate of 75 hours. The table below is an illustration, not observed data, but the shape is the point: each row is a real category of hour that routinely fails to reach an invoice, and the right-hand column shows what the project is paying per hour once you count it.

€6,000 divided by 128 hours is €46.88. The rate you quoted was €80, so the project delivered 41% less per hour than the number you negotiated so carefully. Nothing in that table is a disaster. There is no scope explosion, no client from hell, no unpaid invoice. It is an ordinary, successful, on-time project.

And €46.88 is a pre-tax figure. A US filer owes self-employment tax at 15.3% on 92.35% of net profit before any income tax is calculated; a UK sole trader pays Class 4 National Insurance at 6% on profits between £12,570 and £50,270 on top of income tax. Whatever your effective rate is, it is not what lands in your account.

Why no published average can answer this for you

The instinct at this point is to go looking for a benchmark. Resist it, because the benchmark you want does not exist.

There are exactly two rate datasets in this market with documented methodology, and neither one will help. The freelancermap study above is DACH IT contractors, so unless you are a German-speaking systems engineer it is a number from someone else's market. IPSE's UK average day rates are built on a defensible base of 1,500-plus respondents across SOC major groups 1 to 3, fielded through 2024 with PeoplePerHour, but the figures themselves sit behind a members-only wall, so you cannot check them and neither can anyone quoting them at you.

Everything else you will find is worse. Neither Upwork nor Fiverr publishes a methodologically documented rate-by-discipline table, which means every "average freelance rate for designers in 2026" table circulating online is either a platform's unweighted internal average, a survey of self-selected respondents with no published base, or a blog citing another blog. If a rate table does not tell you who was counted, in which country, in which year, and whether the figure is what the client paid or what the freelancer received, it is decoration.

There is a second reason benchmarks fail here, and it survives even a perfect dataset. A published average is a quoted rate. It is the top line of the table above, not the bottom. Two freelancers quoting an identical €80 can end the year €30 an hour apart, and no survey in the world would detect it.

So the comparison that matters is not you against the market. It is you against you: this project against the last one, this client against your median. That comparison you can actually make, and you are the only person who can.

Measured once is a curiosity, measured every time is a pricing system

Running the calculation on a project you already suspect went badly tells you what you already knew. The value comes from running it on all of them, including the ones that felt fine, because those are where the surprises are.

Two inputs make it possible. The first is capturing hours you have been trained not to capture. Most trackers are built around billable work, so the proposal, the status call and the payment chase never get a timer at all — which means the denominator in your effective rate is always understated and your projects always look better than they were. Worklyn's time tracking treats everything as billable by default and lets you log in plain English after the fact, which is the only way the unglamorous hours ever get recorded. When you are comparing trackers, the useful axis is not stopwatch features but whether tracked hours end up on an invoice, which is the whole basis of the Toggl alternative comparison.

The second input is the fee that actually cleared, net of fees and write-offs, on the date it cleared. Run both through the effective hourly rate calculator: project fee, hours spent, direct costs, and the rate you intended, and it returns the gap. Do it at project close while you still remember the week that went sideways.

What moves the effective rate, and it usually isn't the rate

Look back at the table and note which rows are big. Revisions, delay, unbilled support and chasing account for 32 of the 53 unplanned hours. None of them respond to a price increase. Raise your quoted rate to €95 and the same project produces €7,125 over the same 128 hours, an effective rate of €55.66 — better, but still 41% below the headline, because the leak is proportional.

Scope control moves the number harder than price does. Capping revision rounds at two and pricing the third at your day rate removes 11 hours or bills them. Naming the client's dependencies in the proposal with a dated schedule converts some of those 9 delay hours into either fewer hours or a legitimate change order. Deciding, in writing, that post-delivery support is a separate engagement removes 8 more.

Payment lag is the one people accept as weather. It isn't. IPSE's late payment research finds 35% of self-employed respondents were not paid on time by a client in the past 12 months, with an average amount owed of £5,230, and in a 2022 survey of New York freelancers 76% reported spending one to two hours a week chasing payment. Two hours a week is roughly 100 hours a year of unpaid administrative labor, which is a full project's worth of capacity spent on money you had already earned. The mechanics of removing it belong in the payment system built around time-to-cash, not in your rate card.

Price is the last lever, not the first, and it works best once the others are fixed. If you want to set the quoted rate from your own cost base rather than from a survey of German IT contractors, the rate calculator takes desired annual income, billable hours, weeks off, monthly expenses and tax set-aside, and returns an hourly and day figure. Which of the four pricing models you then wrap around that number is a separate question, and the failure modes of hourly, day rate, fixed price and value pricing decide it.

The threshold at which you stop taking work like this

Measure the effective rate at the close of every project and compare it against two numbers: the rate you quoted, and the floor below which the work does not cover its costs.

Land within 10% of quoted and your scoping is accurate. Price is then the only lever you have left, and raising it will actually stick, because there is no leak to absorb it.

Land 10% to 30% below quoted and the defect is in scope control, not in price. A rate increase gets eaten by the same overrun that caused the gap, so fix the revision cap, the approver count and the free discovery first, then reprice.

Land more than 30% below quoted, or anywhere below your floor, and the conclusion is not a better rate on the next project of that type. It is no next project of that type. The €6,000 example above is a 41% gap, which is the third bucket, and the client would have happily hired you again.

Worklyn's Money on the Table surfaces unbilled hours by project, so the difference between what you worked and what you invoiced stops being something you reconstruct from memory at year end.