DataAnnotation.tech is the most-searched platform in this category and the hardest to get straight information about. The company publishes almost nothing publicly, there is no recruiter to talk to, and the entire entry process is a single unpaid test you get one attempt at.
Two facts decide whether this platform is even relevant to you, and most reviews bury them. First, it accepts contributors from just six countries. Second, the gap between advertised rates and effective earnings is wider here than almost anywhere else. Both are covered below.
Who can join: six countries, no exceptions
DataAnnotation.tech accepts contributors residing in:
- United States
- United Kingdom
- Canada
- Australia
- New Zealand
- Ireland
This is a hard requirement tied to payment processing and compliance, not a preference that persistence overcomes. If you live elsewhere, applying is a waste of your one assessment attempt — go to our guide to AI training jobs outside the US instead, which covers the platforms that genuinely hire globally.
The starter assessment: your single attempt
This is the part that catches people out, so read it carefully.
Entry runs in two stages. The starter assessment takes about an hour and tests writing quality, grammar, reading comprehension and basic logical reasoning — typically rating the quality of short written prompts and answering fact-based questions. If you pass, you move to core qualification tests for specific projects, which use production-difficulty examples and stricter scoring.
The critical detail: the starter assessment is unpaid, and there are no retakes. One attempt, no second chances, and no feedback explaining a rejection. People fail it by skimming instructions or rushing, not by lacking ability.
If you are going to attempt it:
- Block a genuinely uninterrupted hour. This is not a test to take on a phone between other things.
- Read every instruction twice. The assessment is substantially a test of whether you follow detailed guidelines precisely — which is the actual job.
- Write in complete, well-formed prose. Grammar and clarity are scored directly.
- Review your answers before submitting. There is no going back.
Our screening and assessment guide goes deeper on the reasoning patterns these tests reward.
What it pays — advertised vs. actual
The published rates are real, and tiered by track:
- Generalist work — starts around $20–$25/hr.
- Language and multilingual work — $20–$50/hr.
- Specialist tracks — coding, law, mathematics, medicine, physics, finance, accounting, chemistry and biology all start at $50/hr and scale toward $100/hr.
Now the honest part. Contributor-reported effective earnings for generalist work commonly land nearer $14–$20/hr. The gap comes from three places: the unpaid hour of assessment, unpaid time reading project guidelines, and stretches where no work is available in your queue at all.
The specialist tracks hold up much better against their advertised rates, because qualified people are scarcer and projects run longer. If you have a credential that fits one of those tracks, this is a genuinely strong platform. If you do not, calibrate to the lower number. See our market-wide pay breakdown for context.
Payment: the fastest cycle in the category
Payouts are weekly via PayPal, typically landing within 3–7 days. That is among the fastest and most consistent payment cycles of any platform we track, and it is a large part of why the platform's reputation survives its other frictions. Payment reliability is not a complaint people make about DataAnnotation.
You are an independent contractor, so nothing is withheld — our tax guide covers what to set aside.
The work-availability problem
Like Outlier, DataAnnotation routes client project work rather than employing you. Queues empty. Contributors describe strong months followed by weeks with nothing available in their qualified tracks.
The lever you control is qualification breadth: the more core qualifications you pass, the more queues you draw from. Contributors who treat qualifying as ongoing work — rather than a one-time entry cost — report far steadier earnings.
Honest drawbacks
- One-shot entry with no feedback. Fail and you are simply out, with no explanation and no appeal.
- Unpaid assessment time. An hour minimum before you know whether you have access to anything.
- Six countries only. The most restrictive eligibility of any major platform.
- Minimal support. There is no recruiter, no account manager, and communication is thin by design.
- Generalist rates underperform the advertised figure once unpaid time is counted.
Who it is right for
Strong fit: credentialed specialists in the $50–$100 tracks who live in one of the six countries and want fast, reliable weekly PayPal payouts with no interview process.
Reasonable fit: careful writers who follow instructions well, want flexible generalist work, and are calibrated to $14–$20/hr rather than the headline rate.
Poor fit: anyone outside the six eligible countries, anyone needing predictable weekly volume, and anyone who wants a human to talk to when something goes wrong.
How it compares
DataAnnotation pays faster than most and asks less of you up front than Mercor, whose AI-led interview is far more demanding but whose ceiling is higher. It is more restrictive geographically than Outlier but less volatile week to week. Most contributors who do this seriously run two or three platforms at once — the platform comparison lays out which combinations work.
Bottom line
DataAnnotation.tech is legitimate and pays quickly, but it is a narrow door: six countries, one unpaid assessment attempt, and generalist earnings below the advertised rate. If you qualify for a specialist track and live somewhere eligible, it is one of the best options available. Prepare properly for that single attempt — you do not get another one.
Apply at DataAnnotation.tech, or see what is hiring right now on our live job board.