Why we refuse to show results below 5 responses
The single design decision that makes honest answers possible — and why we'd rather show you nothing than a number that outs someone.
Retention
The signals that predict a resignation — and the ones that don't.
Almost every resignation feels like a surprise to the manager and like the most obvious thing in the world to the person leaving. That gap — between "I had no idea" and "I've been unhappy for months" — is where retention is won or lost.
The good news is that people rarely go from content to gone overnight. The weeks before a resignation usually leave a trail. The hard part is that the trail is quiet, and a lot of what looks like a signal is just noise.
What actually shows up first
Across the teams we work with, the earliest reliable signal isn't dramatic. It's a steady drop in one dimension while everything else holds — most often workload or recognition. Not a bad week; a slow slide over three or four pulses.
- Recognition quietly falling. People stop feeling their work is seen well before they stop doing it. This is the most under-watched signal in most teams.
- Workload high and flat. A spike is survivable. A plateau at the top of the scale, pulse after pulse, is what burns people out.
- Comments getting shorter. Disengagement often shows up as less text, not angrier text. When free-text responses thin out, attention has already moved on.
Recognition is the first thing to fall and the last thing managers look at. Flip that order and you catch most of it in time.
The signals that fool you
Just as important is knowing what to ignore, because chasing false alarms burns the trust you need for the real ones.
- A single low score. One bad answer from one person in one week is weather, not climate. This is exactly why results below five responses should never be shown — a single mood is not a trend.
- A dip right after a hard launch. Sentiment falling during crunch is expected and usually recovers. Watch whether it comes back, not whether it dropped.
- The loudest voice. The person who always comments isn't a proxy for the team. Aggregates exist precisely so the quiet majority counts as much as the vocal few.
How to tell them apart
The difference between a signal and noise is almost always direction over time, not the value in any single week. A "6 this week" tells you very little. A dimension that's gone 74 → 68 → 61 → 55 across four pulses tells you a great deal — even if no individual number ever looked alarming.
That's the whole reason to pulse weekly instead of surveying once a year: you're not trying to take one perfect measurement, you're trying to see the slope. Trends are visible in weeks; they're invisible in an annual snapshot.
In PulsePoint: the weekly digest flags exactly this pattern — a dimension sliding across consecutive pulses — and routes it to the right manager before it becomes a resignation. You act on the slope, not the surprise.
What to do when you see it
Catching the signal is only half of it. The response matters more than the detection, and it's simpler than most managers fear: name what you're seeing without naming a person, ask an open question in a one-to-one, and close the loop publicly so the team knows feedback goes somewhere.
You won't save everyone, and you shouldn't try to. But most quiet exits are preventable if you notice the slide while there's still something to talk about. The alternative — finding out in the exit interview — teaches you the same lesson three months too late and one good person short.
Mockup — illustrative editorial content. Written to demonstrate the blog layout, not as validated HR research. Any figures are for design review only.