Skills CX operations
Handle time
Handle time analysis goes wrong in two ways: the average is the wrong statistic for the distribution, and the metric gets pointed at people when it almost always describes work.
npx rulebase-skills install cx-handle-time-analysisWhen to use it
Reach for this when someone says any of these — they are the phrases the skill itself triggers on:
- “why is our AHT increasing”
- “our call handling times are up”
- “which contact reasons take longest”
- “how long do tickets take”
How it works
The method, in the order the skill runs it. The full procedure — tables, worked examples and the edge cases — is in the skill itself.
The average is the wrong statistic
Handle-time distributions are heavily right-skewed and often multi-modal — a dense cluster of routine contacts plus a long tail of investigations. Consequences.
Decompose the time before explaining it
"Handle time" bundles several distinct intervals, and they have different owners: Wait / queue time before an agent engages, Active handling, Hold and research time, Wrap-up / after-contact work, Waiting on someone else.
Concurrency breaks the arithmetic on chat
Chat agents run several conversations at once. Consequences that people miss.
Never rank agents on raw handle time
Handle time is dominated by contact mix, not by the person. An agent with harder work looks slower; one who takes the easy queue looks fast.
What actually moves it
The useful analysis is per contact driver, not per agent. Rank drivers by total time consumed (volume × median time), not by median time alone — a five-minute contact type at high volume usually outranks a two-hour rarity, and the ranking by median points you at the wrong work.
Related skills
Free and open source, and vendor-neutral — it reads the conversations from whichever helpdesk you already run. Browse all 149 skills · connect your helpdesk over MCP · source on GitHub
