The most common reasons task submissions are rejected

Almost all rejections come from the same handful of causes. None of them require more talent to avoid, only more attention.

Not following the brief

The largest category by a wide margin. Someone skimmed the brief, recognised the general shape of the task, and answered from memory of a similar one. Read the whole brief every time, including tasks that look identical to the last twenty.

Wrong length

Answers far shorter or longer than specified. If a range is given, land inside it. If none is given, answer completely and stop.

Wrong language

Answering in English when the brief asked for a specific language, or mixing languages within one answer where the brief asked for one. This is a common and entirely avoidable rejection.

Generic content

Answers that could apply to anything. Reviewers are collecting specifics, and a submission with no concrete detail has no value to the dataset even if every sentence is grammatical.

Generated text

Text produced by a model, submitted for a task that asked for human writing. It is recognisable, it defeats the purpose of the task, and it puts your account at risk. Write your own answers.

Poor media quality

Blurred photos, dark images, cropped edges, background noise on a recording, or a file that does not match the specification for lighting, angle or framing. Check the file before uploading, not after.

Duplicate or copied submissions

Reusing your own previous answer across tasks, or submitting someone else's work. Both are detected and both carry consequences beyond the single rejection.

Incomplete submissions

Missing a required part of a multi-part task. Scroll to the end of the form before submitting.

What to do after a rejection

Read the reason, apply it to the next task, and move on. One rejection is noise. A pattern of the same rejection reason is a signal, and contributors who act on that signal see acceptance rates rise within a handful of tasks.