Microtask

What counts as a microtask

A microtask is a small, self-contained unit of work — labeling one image, transcribing one sentence, rating one search result — that takes seconds to a few minutes and pays cents. Nothing about the task requires context beyond its instructions, which is exactly why anyone can start.

The most common kinds are annotation work — boxes around objects, tags on text — plus short transcription snippets and search-result ratings. Tasks arrive in a queue, and you work through as many as your session allows, one accepted unit at a time. Platforms often batch related items into short sessions, so the queue you see at nine in the morning may not exist by noon.

• Labeling a single image or video frame
• Tagging entities or categories in a short text
• Transcribing a few seconds of audio
• Rating the relevance of one search result

How microtask pay works

Microtasks pay per accepted item, usually cents each, under a per-task pay model: the price is shown before you start, and rejected work typically pays nothing. Hourly income only materializes at volume — hundreds of tasks, done quickly, with few rejections.

That is why this directory judges microtask platforms on effective hourly rate rather than task prices: three cents per label is meaningless until you know how many labels fit into an hour. Track both numbers — tasks per hour and cents per task — before deciding where the hour goes.

Microtasks versus longer AI-training projects

Microtasks are the no-application end of AI training work: sign up, pass minimal checks, and start. Longer projects — guideline-heavy annotation campaigns, expert evaluation, search-quality contracts — pay more per hour but screen harder and can take days to onboard.

The trade-off is the entry barrier itself. Microtasks suit absolute beginners who want to earn something today and learn how guidelines and quality checks work; they produce the lowest and most variable effective rates in the directory, so move up as soon as your accuracy does.

How to make microtask work worth your time

Worth comes from three habits: speed without sloppiness, reading guidelines before touching tasks, and working when the queue is full. Availability swings by day and region, so an empty Tuesday queue is a demand signal, not a verdict on the platform.

Track your own rate per hour for a week, then compare platform pages in the AI-training hub, where pay models and availability notes are listed per platform. If your measured rate stays below your floor, those hours are better spent on higher-paying task types.