# Paid Test Project for Freelancers: Scope, Pay, Rights

> A paid test project is a small piece of real work, typically two to six hours, that you pay for at the freelancer's normal rate before committing to a larger engagement. Agree the scope, time cap, payment, delivery format and ownership of the result in writing, and only offer it to your two or three strongest candidates.

- URL: https://freelancerbewertung.de/en/vetting/paid-test-project/
- Language: en
- Updated: 2026-10-19
- Publisher: Freelancerbewertung (OnlineBuilders LLC, https://onlinebuilders.biz)

Experienced freelancers often turn down unpaid test tasks. They are booked, and a free afternoon for a stranger is a bad trade. That means an unpaid test quietly filters out the people you most want to hire.

So pay for it. The case is commercial rather than ethical, and the numbers are small.

## Why a test beats every other vetting step

Portfolios and references describe the past. The test shows how someone deals with your material, your gaps in the brief and your deadline. It is also the only step where you compare candidates on identical work, which is why it carries the most weight in our scorecard on [how to vet a freelancer](https://freelancerbewertung.de/en/vetting/how-to-vet-a-freelancer/).

What you learn goes beyond the output. You see whether they ask questions before starting, whether they flag problems or hide them, and how they hand over work. Those habits will repeat on the real project.

## What a paid test costs, and what it saves

Take a freelance developer at $100 an hour, a mid-range figure from our overview of [freelancer hourly rates](https://freelancerbewertung.de/en/contracts/freelancer-hourly-rates/). Compare a short test for two finalists with the cost of a bad hire discovered three weeks in.

**Example: cost of testing vs cost of a wrong hire**

- Test: 2 candidates × 4 hours × $100: $800
- Wrong hire: 3 weeks × 25 hours × $100: $7,500
- **Difference in favour of testing: $6,700**

Our worked example at $100 per hour. The wrong-hire figure ignores delay and rework, so the real gap is larger.

The calculation leaves out the expensive part: the weeks lost before you start again with someone else.

## How big should the test be?

Two to six hours is the useful range. Shorter, and you learn little more than you would from a conversation. Longer, and it has become a small project. At that point, call it a first milestone and contract it properly.

Make it real but not critical. Pick a self-contained slice of your own backlog rather than an invented exercise, because then you can judge the result against your own standards.

| Role | Good test | Poor test |
|---|---|---|
| Backend developer | Fix a documented bug in a copy of the repo, with a test | Algorithm puzzles unrelated to your stack |
| Frontend developer | Build one component from your existing design system | A full landing page |
| Designer | One page in your current visual style, plus two variations | A new logo "just to see" |
| Writer, SEO | Outline and intro for a planned article | Five finished articles |
| Paid ads specialist | Audit of an existing account with five prioritised fixes | Launching a live campaign with real budget |

Keep real customer data out of it. If a contractor processes personal data on your behalf and GDPR or UK GDPR applies, you need a processing agreement under Article 28. Anonymised or synthetic test data avoids that entirely.

## Who owns the result?

This catches people out. Paying for work does not, by itself, make you the owner of the copyright.

US law treats a contractor's work as "work made for hire" only if it falls into one of the specific categories listed in [17 U.S.C. § 101](https://www.copyright.gov/title17/92chap1.html#101) and both parties sign a written agreement saying so. Most software and many design jobs do not fit those categories, so you need a written assignment instead. British law requires an assignment of copyright to be in writing and signed by the person assigning it.

If you only want to evaluate the test, none of this matters. If you might ship the result, add a line to the agreement. Our guide on [who owns the copyright when you hire a freelancer](https://freelancerbewertung.de/en/contracts/freelancer-copyright-ownership/) covers the wording in more detail.

## A test brief you can copy

Vague briefs produce results you cannot compare. Then you are not measuring who is better, only who guessed better.

**Test project brief**

- [ ] Goal in one sentence: what exists at the end?
- [ ] Materials: access, files, style guide, sample data
- [ ] Time cap in hours and delivery date
- [ ] Delivery format, e.g. pull request, Figma link, shared doc
- [ ] Acceptance criteria you will judge against
- [ ] Payment: hourly rate and maximum hours
- [ ] Ownership of the result if you may use it
- [ ] Who to contact with questions during the task

A short email that covers these points is enough. Ask the candidate to reply confirming the rate and cap before they start. If you already wrote a solid [freelance job post](https://freelancerbewertung.de/en/finding-freelancers/freelance-job-post/), half the brief is done.

## Running the test without wasting anyone's time

Most of the friction in test projects is logistics. Prepare access before the start date: a repo copy, a staging login, a design file with edit rights. If a candidate spends the first hour waiting for credentials, that burns your money and their goodwill.

Say when you are available for questions, for example a two-hour window on the first afternoon. Then actually be there. How quickly you answer affects the result, and you want to compare candidates, not your own response times.

Pay within a week of delivery, whatever the outcome. The freelancers you reject today may be the right fit for the next project, and they talk to each other. Two sentences of honest feedback cost you five minutes and leave a better impression than silence.

## Scoring the result

Rate each point from 1 to 5, immediately after delivery:

- Does the output meet the acceptance criteria?
- Were sensible questions asked before work began?
- Was the time cap respected, or raised early?
- Is there a short note on what was done and what is still open?
- Were assumptions stated where the brief had gaps?

That last point often separates professionals from everyone else. A note like "I assumed the API paginates, worth confirming" will save you real trouble once the project is running.

## When to skip the test

Sometimes it adds little. Referrals from someone who has run several projects with the freelancer give you most of the same evidence. Very senior consultants may find a test odd. Either way, a short first milestone with its own sign-off date does the job under a different label.

The same goes for very small jobs. If the whole engagement is six hours of work, the test would be the project. Hire, set a clear acceptance criterion and treat the first delivery as the evaluation. Losing a few hundred euros or dollars on a bad pick is cheaper than running a selection process around it.

And if the test shows your project is far bigger than you thought, that is useful too. For larger web applications, a team that already works together, such as [OnlineBuilders](https://onlinebuilders.biz/solutions/software-development), can be a better fit than one freelancer.

Your next step: find a task in your backlog that you could finish yourself in half a day and judge precisely. Write the brief with the checklist above and send it to your top two candidates this week.

## Step by step

1. **Pick a real, self-contained task** Choose a piece of your project that fits into two to six hours and that you can judge yourself.
2. **Write the brief** State the goal, materials, time cap, delivery format and acceptance criteria.
3. **Agree pay and ownership** Confirm rate, hour cap and who owns the result by email before work starts.
4. **Score the result** Rate the output, questions asked, time discipline and handover notes on a fixed scale.

## FAQ

### Is it acceptable to ask for an unpaid test?

It is legal in most places if both sides agree, but it is a poor filter. Busy, experienced freelancers tend to say no, so you end up testing the people with spare time. A short paid task costs little and attracts better candidates.

### Do I own the work from a paid test?

Not automatically. In the US and UK, copyright in a contractor's work stays with the contractor unless there is a signed written assignment. Add a one-line ownership clause to the test agreement if you might use the result.

### What if a candidate refuses any test at all?

Some senior specialists prefer not to do tests. Offer a short first milestone with its own sign-off date instead. It tests the same things under a different name.

### How do I compare test results fairly?

Give every candidate the identical brief, materials and time cap, and score them on the same points straight after delivery. Change the brief between candidates and you are comparing guesses.

Photo: Tamarcus Brown tamarcusbrown, CC0 (https://commons.wikimedia.org/wiki/File:Precise_drawing_(Unsplash).jpg)