Every warning and ban we issue is weighed against the full context of what happened: the report, the user involved, what came before. We move quickly on enforcement because at an in-person event, speed is what stops harm from happening twice. But fast isn't the same as certain. Sometimes the person on the other end of a decision thinks we got it wrong, and sometimes they're right.
Opting for permanent, unappealable decisions might streamline operations, but it would also make us less accountable, and no more accurate. So we built appeals into enforcement from the start rather than adding them later.
An appeal is a check, not a second chance
An appeal doesn't reopen the case. Our Community Guidelines put it plainly: an appeal is a check that we got it right, not a second chance. If you think we identified the wrong person, or misread what happened, that's what an appeal is for.
That framing matters to us because it keeps the process honest in both directions. Sometimes we get a call wrong, and a member ends up facing an enforcement action they don't deserve. The appeal exists so we can catch that and fix it.
But when we get a call right, it should stay right. The appeal is a chance to bring something we missed. Sending the same request five times doesn't change the outcome, and it shouldn't. Personally, that's the part I care about most, because of the guests who reported in the first place. Reporting someone after a dinner takes effort, and sometimes a bit of courage. If pushing hard enough could undo that, I think a lot of people would stop reporting.
Fresh eyes, and three possible outcomes
Every appeal is reviewed by someone who wasn't involved in the original decision. Appeals sit in their own queue, separate from the reports that led to the enforcement action, and the reviewer goes back to the original evidence. If the decision stands, we tell the member why.
Upheld means the original decision stands. Overturned means the account comes back as it was: profile, connections, events and messages. Modified means the reviewer agrees something happened, but the user provided additional evidence or information that changed the picture. Most often that's a permanent ban reduced to a formal warning, because the new information showed the incident was real but less serious than it first seemed.
The ladder, and the exceptions to it
Most violations follow a ladder. A first, low-severity issue gets an educational nudge. A repeat gets a formal warning, and if it continues or escalates, a ban. Appealing a decision on the ladder means asking if it was assigned to the correct level.
Some more serious violations skip the ladder and get a ban on the first offence. The user can still appeal one, but the reviewer won't be weighing severity, since zero tolerance already sets it. They're checking the same two things as always: that we have the right person, and that we read the situation correctly.
What the numbers say so far
Over the past six months, about 10% of enforcement actions were appealed. Of the appeals we've closed so far, we upheld close to 90%. The rest were either overturned or modified, and modification is the least common of the three outcomes.
We also aim to review every appeal within 72 hours, and most get a decision within a few days. Speed matters because a ban we got wrong costs the member something every day it stays in place. They can't book, they lose sight of upcoming events, and they may miss plans they'd already made. When we overturn a decision, we want them back in the app as quickly as possible.
These are early numbers, and our tracking is still catching up, so expect them to move. But the shape is the one we'd hope for: reviewers who re-read each case, usually agree with the first call, and reverse or adjust about one decision in ten.
Every correction teaches us something
When an appeal is overturned or modified, the member gets a correction and we learn that something earlier in the process was off. It could be a reviewer’s judgement, or one of the systems that flag cases for review in the first place. Either way it shows us where enforcement needs recalibrating, and without appeals we'd have no way of seeing it.
This is still a work in progress. Our tracking is getting better, and so is the process behind it. But we'd rather be judged on how willing we are to check our own decisions than on how rarely anyone asks us to.
Nolwenn, Trust & Safety Manager at Timeleft
