Trust report
A grade for every number, and the formula behind it
Plaintrace measures what most analytics hides: how much traffic it lost, how much of what it got was not human, and how much of the rest never interacted. Then it turns that into a grade you can recompute.
The formula
Published, so you can check it
The score starts at 100 and loses points for each kind of doubt. A is 85 and up, B 70, C 50, below that D.
score = 100 − 40 × loss
− 30 × non-human share
− 20 × zero-interaction share
− 10 × refused / (received + refused)
A ≥ 85 · B ≥ 70 · C ≥ 50 · D belowMeasured loss
weight 40Page loads your server answered (counted by the server SDK) that never arrived as a tracked page view: ad blockers, privacy browsers, failed scripts. Not measured without the server SDK, and the report says so.
Non-human share
weight 30Events scored as bot or suspicious: known bot user agents, headless browsers, datacenter networks, inhumanly fast sequences.
Zero-interaction visits
weight 20Visits with no active time at all: opened, never interacted with. A high share usually means prefetching, link previews or bots that look human.
Refused events
weight 10Events the collector refused after reading them (invalid names or addresses, too old, heatmaps off), as a share of all events sent.
- Human traffic
- 98.2%
- Measured loss
- 4.1%
- Zero-interaction visits
- 5.3%
bots and suspicious traffic kept apart, not deleted
12,986 page loads served, 12,453 tracked
opened, never touched
100 − 40·loss − 30·non-human − 20·zero-interaction − 10·refused
What it changes
Decisions that account for doubt
Small samples are flagged
Honest visitor counts
Bots stay visible
See your own grade.
Free for two sites and 100,000 events a month. Every feature included, no card needed.