How to work out your real hourly rate on AI tasking platforms

DataAnnotation, Outlier, Alignerr, Mercor and the rest all advertise an hourly rate. Almost nobody doing the work earns that number. Here is the honest calculation, a worked example, and the traps that hide the difference.

The only formula that matters

Effective rate = money that actually landed ÷ every hour you actually spent.

Every hour means all of it: reading rubrics and onboarding docs, doing tasks that were later rejected, redoing flagged work, qualification tests, and time spent hunting for available tasks. If the hour existed because of the platform, it goes in the denominator.

A worked example

Say a platform advertises $25/hr and in one week you:

• spend 2 hours on rubric reading and a qualification test (unpaid)
• complete 10 tasks logged at 8 billable hours
• have 2 of those tasks rejected (1.5 of the 8 hours, unpaid)
• spend 1 hour redoing a flagged task (unpaid)

Paid: 6.5 accepted hours × $25 = $162.50. Spent: 2 + 8 + 1 = 11 hours. Effective rate: $14.77/hr — 59% of the advertised number. Not necessarily a bad rate, but a very different decision than the one the headline suggested. And it can swing the other way: efficient workers on generous projects sometimes beat the advertised rate. The point is to know, not to guess.

The three traps

1. Counting promised money as earned. Platforms pay per accepted task, on a lag. Until it lands, it is exposure, not income. Track "submitted but unpaid" as its own number — if a platform goes quiet, that number is what you are risking.

2. Forgetting the unpaid hours. Memory is generous. A timer is not. Log time as it happens, including the boring parts.

3. Averaging across platforms. One platform at $19/hr effective and another at $8/hr effective average to a meaningless number. Track per platform and drop the ones below your floor.

Set a floor before you start

Decide the rate below which the work is not worth your time, and check your weekly effective rate against it. Two bad weeks in a row is information; act on it rather than hoping the next project pays better.