5 Things I Wish I Knew
Before My First Task

Most of the content on this site is organized by platform, pay tier, or audience type. This piece is different — it's the personal, first-person list I'd actually hand to someone starting today, drawn from 14 months and €18,000+ earned across Mercor, SME Careers, micro1, and others.

1. Reading the Guidelines Is the Whole Job

This sounds obvious and yet it's the single most common way beginners underperform in their first weeks. AI training task quality is measured almost entirely on guideline adherence — not raw intelligence, not writing skill, not how fast you are. If you skim the instructions and rely on common sense to fill in the gaps, you will consistently miss platform-specific requirements that common sense would never predict.

Before my first task on DataAnnotation.tech, I read the guidelines twice because they were long. I passed the assessment on the first attempt. That was not a coincidence.

2. Month One Income Is Not Representative

I made less in my first month than I expected, and I almost quit. What I didn't understand then — and what I explain in our income report — is that month one is structurally the lowest-earning period for almost everyone. You're still completing assessments, building track record, and getting matched to tasks for the first time. Month three looked meaningfully different from month one. Don't judge the whole category by your first few weeks.

3. Apply to Multiple Platforms Before You Hear Back From the First

I wasted three weeks waiting on a single platform's response before applying anywhere else. Apply to two or three platforms simultaneously from day one — Mercor, SME Careers, and micro1 were my starting stack — and let the fastest one become your first actual earning platform rather than losing weeks to a single slow response. This is the entire argument behind our income stack guide.

Waiting for one response before applying to the next is the most expensive mistake a new AI trainer can make in terms of time wasted during the startup period.

4. Your Hourly Rate Depends Heavily on Task Selection, Not Just Platform

The pay range on any given platform is much wider than it first appears, because the same platform can have tasks ranging from $15/hr effective to $60+/hr effective depending on task complexity, your accuracy, and how long you take. Early on, I tracked this poorly and assumed a platform was "low paying" when I was actually just choosing slow, lower-paying tasks. Tracking your actual hourly rate per task type — not just overall monthly income — shows you which work is genuinely worth your time. This connects directly to what we describe in the pay gap explainer.

5. Check Your Spam Folder Before Assuming Rejection

I missed a full week of available tasks on DataAnnotation.tech because their acceptance email went to spam. I'm not alone in this — it's mentioned in our timeline guide and in forum discussions across this space. Check spam specifically after applying to any platform, ideally daily for the first two weeks. This is embarrassingly simple and genuinely costly when missed.

The One Thing That Matters Most

If I had to compress all of the above into a single sentence: treat the application and onboarding period as seriously as you'd treat the first week of a real job, because how you perform during it shapes everything that comes after — consistent with what we now describe in our piece on how onboarding effort shapes long-term task matching.

For a complete overview of how the reviewing process works, see our get paid to review AI guide.