Methodology

Where every number on this site comes from, and what it can and can't tell you.

Sources

Once a day we read each platform's public job listings, the same pages anyone can open without logging in, and only pages their robots.txt allows. Last run: October 19, 2026 at 2:25 AM ET.

Platform	What we read	Last run
Mercor	The public explore page at work.mercor.com	300 roles
micro1	Roles and pay from micro1's referral program listings, checked by hand at least weekly	24 roles
Handshake AI	Opportunity pages listed in joinhandshake.com's public sitemap	242 roles
Alignerr	The public job list at alignerr.com/jobs	528 roles

Outlier and DataAnnotation show projects only after login, so they have profiles but no live listings.

How pay is normalized

Hourly ranges are used as listed. A single figure (“up to $120/hr”) counts as both the low and high end.
Salaried roles are converted at 2,080 hours a year, and labeled as salaries.
Per-task and one-time pay are shown as listed but left out of hourly medians: there's no honest hourly figure for them.
Medians use the midpoint of each hourly range.

Grouping and fields

Alignerr posts many near-identical copies of a role, one per country. We group listings with the same title and pay into one role and show how many location versions exist. Fields (medicine, law, languages and so on) are assigned from each title by rules we test; if a role looks misfiled, tell us.

What this can't tell you

Listings show what platforms advertise. They don't show how many hours you'll get, how often projects pause, or what reviewers approve. That's why worker reports exist, and why every platform profile cites its claims.

Independence

We aren't affiliated with any platform here. Some pay referral bonuses; that never changes rankings or what profiles say. See how we make money.

Platforms covered: Mercor, micro1, Handshake AI, Alignerr, Outlier, DataAnnotation, Prolific, Invisible, Turing, Remotasks, Appen, TELUS Digital AI Community, Toloka, OpenTrain AI, Braintrust.