What the score actually measures
Your trust score is a single number between 0 and 100 that reflects the share of your responses approved across the whole platform. The formula is deliberately simple and impossible to game by volume:
“Trust score = approved responses ÷ total responses × 100”
Because it is a ratio, submitting more responses does not raise it - submitting better responses does. A participant with 19 of 20 approved (95) outranks one with 60 of 100 approved (60), even though the second submitted five times as many.
The bands and what they unlock
| Score | Band | What happens |
| 80-100 | Premium | Fast-tracked rewards, priority invitations, access to high-paying B2B and specialist studies |
| 60-79 | Standard | Normal processing on all consumer studies |
| 30-59 | Watch | Rewards may be held while a researcher reviews the response |
| 0-29 | Flagged | Every future response is auto-flagged for manual review; withdrawals paused |
How to raise it
- Take roughly the estimated time the survey shows - the speed check flags completions under 20% of the estimate
- Read every question; attention checks are injected at random and failing one is the strongest single negative signal
- Vary rating answers when your opinion actually varies - identical answers on 85%+ of ratings triggers the straight-line detector
- Write at least one full sentence on open-text questions
- Never share an account or connect through a VPN to enter surveys targeted at another country
How fast does it recover?
The score has no memory decay - it is recalculated from your full history after every completion. Early mistakes fade as approved responses accumulate: a participant who failed 2 of their first 5 surveys (score 60) reaches 90 after 25 consecutive approvals. The math always allows a comeback; it just rewards consistency over time.
“Trust is earned in drops and lost in buckets.”
- An old saying that also describes our scoring
Want the research behind the system? We published a study showing trust scores predict next-response quality with an AUC of 0.83 - better than any single per-survey check.