Metrics reference - Plain Docs

For who gets credit for each metric, see team reporting. Each metric links to a page with the full calculation. All of them use the exclusions every metric shares and can be filtered and grouped.

Workspace metrics

Metric	What it measures
Threads in Todo	Threads in Todo at the end of each bucket
New threads created	Threads created
Threads re-opened	Threads that moved from Done back to Todo
Threads moved to Done	Threads that moved to Done
First response time	First inbound customer message to the first reply. Threads whose first reply came from a machine user are left out
Next response time	Each later customer message to the next reply
Resolution time	First inbound customer message to Done
Customer waiting time	Total time in Todo or Paused for later
SLA compliance: first response	Share of threads whose first response SLA was achieved
SLA compliance: next response	Share of threads whose next response SLA was achieved
CSAT: % positive	Share of survey ratings that were Positive
CSAT responses	Survey ratings, split into Positive, Neutral, and Negative
Heatmap: threads created	Threads created, by hour and day of week

Count metrics count each thread once per bucket. A thread that moves to Done twice in the same day counts once in that day.

Team member metrics

These appear only on a team member’s page in team reporting.

Metric	What it measures
Messages sent	Outbound messages the user sent
Messages sent by tier	Outbound messages the user sent, split by the thread’s tier
Thread assignments	Threads assigned to the user. A thread assigned twice in one bucket counts once
Heatmap: messages sent	Threads the user sent a message on, by hour and day of week

Percentiles

Duration metrics show the median and the 90th percentile:

Median: half of the measurements were faster, half slower. Outliers move it less than they move an average
90th percentile: 90% of the measurements were faster than this value. Use it to watch the slowest 10%

Charts grouped by a dimension show the median only. The Reporting API accepts any percentile from 1 to 99.