How the score works
What the Productivity Score measures, and where it falls short.
What the score is
The formula: share of active tracked time spent in apps or sites classified Productive. Neutral and Distraction count against it equally; idle time doesn't count.
Same formula everywhere: a company or team score sums everyone's tracked samples first, then takes the share, so more tracked time weighs more — not an average of individual scores. An individual score is the same formula over their own samples alone.
Worked example. An 8-hour shift: 6.5 hours active tracked time, 1.5 idle (60+ seconds with no input), dropped rather than counted as zero. Of the 6.5 active hours: 5.0h Productive, 1.0h Neutral, 0.5h Distraction. Score = 5.0 ÷ 6.5 × 100 = 77.
The bands
Three colors, two cut-offs:
| Band | Score | Meaning |
|---|---|---|
| Green | 70–100 | Most active time was classified Productive. |
| Amber | 40–69 | A mix, often Neutral-heavy rather than actively unproductive. |
| Red | 0–39 | Most active time was Neutral, Distraction, or both. |
77 lands in green. Swap the mix — 1.0h Productive, 4.5h Neutral, 1.0h Distraction, a meeting-heavy day — and the score is 15: solidly red, despite zero goofing off. See Honest limits, below.
Where classifications come from
Every sample, one executable or browser host, captured every 10 seconds by default, resolves to Productive, Neutral, or Distraction by fixed precedence:
- A rule you've added for this tenant, an executable or, for browser activity, a site host, set under Classifications → Classified → Add a rule. Overrides everything below, including our own catalog.
- For browser activity: our curated site catalog, consensus across SnitchOS on that host, then a general content-category list limited to its highest-confidence categories (adult, gambling, games, social networking, and similar).
- Our curated app catalog's default for that executable.
- Consensus across SnitchOS again, for an unknown executable.
- Still nothing: Neutral.
Consensus: three or more organizations independently classify the same site or executable; at 80%+ agreement, that verdict fills gaps neither a tenant rule nor the catalog has an opinion on — never outvoting either.
Add or change a rule from Classifications → Classified, or work the Pending queue. Deleting one reverts every event, past and future, to whatever the catalog, consensus, or content list would say without it.
Strict Mode changes the denominator
The Dashboard and Working Hours report carry a Strict Mode toggle, off by default; it moves every score at once.
Normally a sample counts active whenever the device isn't idle: foregrounded app, idle timer not tripped. Strict Mode also requires a keystroke or click in that interval — samples with no input fold into idle instead.
Nothing is re-classified, nothing extra collected — the toggle decides which stored samples reach the calculation; score moves either way, up when quiet time was Neutral, down when Productive. Active/Passive columns in Working Hours Detailed stay raw, unaffected.
What doesn't move the number
- Idle time. Dropped from the denominator, per the example above. Pure system chrome (lock screen, Start menu, installer noise) drops the same way.
- Keystrokes, clicks, mouse distance. Logged as activity counters elsewhere; the score has no input-count term, only which category a sample falls into.
- Screenshots. Stored separately for spot review, never read by anything that computes a score.
Honest limits
A day of legitimate Neutral work — research, spec reading, an unfamiliar tool — scores the same as a day of scrolling; the arithmetic can't tell them apart, only a person can.
The score measures a person against their own tracked time, not a teammate's — no percentile, no curve, unaffected by Dashboard ranking. Adherence compares hours worked to the shift schedule — a different question.