LearnAI home

Getting Better Work ยท Lesson 6

Your First 30 Days in Annotation Work

A realistic plan for starting out, plus the Label Maker badge.

A plan built for how this actually goes

Most people start annotation work by signing up to one platform, waiting, and losing interest. The approach below is designed around the two things that reliably happen instead: most applications go nowhere, and the work arrives in unpredictable bursts.

Everything here is built outside this site. There is no account to create and nothing is stored anywhere. Copy the blocks into a notebook, a document or a spreadsheet you control.

In plain English

Onboarding:
Everything between being accepted and doing paid work: paperwork, guidelines, tests and platform setup.
Payment cycle:
The schedule on which a platform releases money, for example monthly in arrears rather than on completion.
Application tracker:
A simple list of where you applied, when, and what happened, so you follow up rather than forget.
Burn rate:
The unpaid hours you spend before any money arrives. Worth knowing before you commit them.

Week one: apply broadly

Send eight to twelve applications, not one. Response rates in this industry are low and slow, and the cost of applying to a tenth platform is twenty minutes.

Run every one through the five tests from lesson three before you create an account, and use the enquiry message to ask about pay structure, payment timing and content type. Keep a tracker so that when a reply arrives three weeks later you remember which application it belongs to.

Application tracker
DATECOMPANYROLEHOW I FOUND IT5 TESTS PASSED?STATUSNEXT STEP

STATUS: applied / assessment sent / passed / rejected / silent / withdrawn Follow up once after 14 days. Once only.

Week two: expect assessments, and expect to fail some

Qualification tests are normal, usually unpaid, and often harder than the work itself because they concentrate the edge cases. Failing one is ordinary and says very little about you. Some platforms let you retry after a waiting period, some do not.

Read the guidelines before the test rather than during it, and take notes in your own words. That is also the exact habit that protects your quality score later.

Do not quit a job, turn down other work, or pause your wider job search for this. Nothing is confirmed until money has actually reached your bank account, and even accepted applications can produce no available tasks for weeks.

Week three: work, and measure

Once tasks appear, start logging from the very first one. The point is not the money in week three. The point is the data that tells you whether week ten is worth having.

Daily work log and rate tracker
DATEPLATFORMPAID MINSUNPAID MINSTASKS DONEEARNEDREJECTEDNOTES

UNPAID MINS includes: guidelines, tests, waiting for the queue, appeals, invoicing, chasing payment.

WEEKLY TOTAL Earned after fees = ____ Total minutes (paid+unpaid) = ____ EFFECTIVE HOURLY RATE = earned divided by (minutes divided by 60)

Log the boring things too: which days the queue was empty, how long support took to reply, whether a promised payment date held. Those patterns matter more than any single good day.

Your first payment may well arrive after day 30. Payment thresholds and monthly cycles mean the gap between doing the work and holding the money is commonly several weeks. Never plan a bill around money you have not received.

Week four: review honestly

At day 30 you sit down with your own numbers and answer five questions without flattering yourself.

The day 30 review
  1. What is my effective hourly rate, after fees and unpaid time?
  2. How does that compare to the minimum wage where I live, and to anything else I could do with those hours?
  3. How many hours of work were actually available per week?
  4. Has any money reached my bank account yet? If not, when will it?
  5. How do I feel after a three hour session? Fine, drained, or worse?

THEN CHOOSE ONE Continue the rate and the hours justify the time Narrow drop the weakest platforms, keep the best one or two Specialise apply for language or domain work using lesson five Stop the numbers do not work, and stopping is a decision rather than a failure

Stopping is a legitimate outcome. If your tracker shows a poor rate, empty queues and no payment yet, you have not wasted the month. You have bought a clear answer for the price of some unpaid hours, and you can now put your energy somewhere better instead of spending six more months finding out slowly.

Checkpoint

Apply broadly, expect unpaid assessments, log every minute paid and unpaid, keep every other income source, and decide at day 30 using your own numbers.

The whole 30 days on one page

WEEK 1 [ ] 8-12 applications sent [ ] All five platform tests run on each [ ] Content type and pay structure asked about in writing [ ] Tracker started

WEEK 2 [ ] Guidelines read before any assessment [ ] Notes written in my own words [ ] No job quit, no other application withdrawn

WEEK 3 [ ] Logging from the very first task [ ] Unpaid minutes recorded honestly [ ] Queue availability and support response times noted

WEEK 4 [ ] Effective hourly rate calculated [ ] Payment status confirmed [ ] One decision made: continue, narrow, specialise or stop

NEVER [ ] Paid anyone for training, equipment or a starter kit [ ] Sent ID or bank credentials before a contract [ ] Forwarded parcels or money for an employer

What you have

You now know what the work is, how the pay is really structured, how to judge a platform, how to recognise the scams that target people looking for work, and what would move you towards better-paid projects. That is a realistic picture, which is more than most of what is written about this subject offers.

Finish the quiz and the Label Maker badge is yours. Keep the tracker either way. It will still be telling you the truth in month three, when the adverts have stopped being interesting.

๐Ÿ“ Quiz

Question 1 of 4

Why does the plan recommend eight to twelve applications rather than one or two?

Found this useful? Pass it on.