Reviewing Work
The approval workflow for evaluating participant submissions, including manual review, auto-judge, and quality controls.
Reviewing submissions is how you ensure the quality of participant work. Terac provides multiple review methods to match different opportunity needs.
Review Methods
| Method | Description | Best For |
|---|---|---|
| Manual review | You review each submission individually | Complex or nuanced work |
| Auto-judge | AI evaluates submissions against criteria | Standardized tasks with clear pass/fail criteria |
| Quality check | AI screens first, you review edge cases | High-volume opportunities needing efficiency with quality |
Manual Review Workflow
Open the submission
Navigate to the Submissions tab and click a submission in the "Awaiting Review" state.
Review the work
Examine the participant's submission:
- Survey responses
- Uploaded files
- Interview recordings and transcripts
Make a decision
Choose one of three actions:
- Approve - the work meets your standards; the participant is paid
- Reject - the work does not meet standards; optionally provide a reason
- Strike - flag serious quality issues (repeated low effort, dishonesty)
Screening Review
If your opportunity uses manual screening (no auto-grading rules), screening responses also land in a review queue. For each applicant you can:
- Approve screening - the participant advances to the task
- Reject screening - the participant is screened out
Auto-Judge
The auto-judge uses AI to evaluate submissions against criteria you define. It works well for tasks where quality can be assessed programmatically:
- Completeness checks (all required fields filled)
- Content quality evaluation
- Compliance with instructions
The auto-judge assigns a pass or fail recommendation. You can configure whether to auto-approve passes or route all results through manual review.
Auto-Review for External Tasks
When your task runs on a third-party tool and you have configured redirect URLs, submissions are reviewed automatically from the result your tool sends back:
- A
completedcallback approves the submission and triggers the payout. - A
screened_out,quota_full, orrejectedcallback closes the submission without charging you.
If redirect URLs are enabled but no callback arrives, the submission stays in Awaiting Review and is auto-rejected after 6 hours. This is why configuring the completion redirect matters: without it, qualified participants are rejected automatically.
Participant Ratings
After reviewing a submission, you can rate the participant on a quality scale. Ratings are visible in the participant's profile and influence their standing for future opportunities. High-quality participants are more likely to be recommended for your next opportunity.
Kick Out
If a participant is problematic (unresponsive, submitting spam), you can remove them from the opportunity using the Kick Out action. This cancels their submission and frees the participant slot.
Reset Participation
In rare cases where a participant needs to restart (e.g., technical issues during a task), you can reset their participation. This returns their submission to the beginning of the task phase.
Resetting participation cannot be undone. Only use this when a genuine technical issue prevented the participant from completing their work fairly.