Settingsintermediate

Metric types reference

First Reply Time, All Messages, and Resolution Time — what each measures, when each starts and stops, and how to combine them across policies.

4 min read

Metric types reference

Each policy tracks one of three metrics, chosen under Metric to measure. Create one policy per metric you’re committing to.

First Reply Time

Measures how long until the agent’s first response after the customer’s initial message.

  • Starts — When the conversation is created (first inbound customer message arrives)
  • Stops — When the first reply from an agent on your team, or an answer from your Agent Stack, is sent
  • Restarts on reopen? — Only if nobody had replied before the conversation was closed. Once a reply has stopped it, this metric never restarts and keeps the Met or Breached result of that reply.
  • Best for — Channels where first impressions matter — email, chat, contact form

Only a real reply stops the timer: a reply from an agent on your team, or an answer from your Agent Stack. An automatic reply that an automation sends, an internal note, or a system message does not stop the timer and does not count as the first reply.

This is the most-tracked SLA metric in customer support. Customers judge the quality of support partly by how fast someone responds to their problem, separate from how fast it gets resolved.

All Messages

Measures how long until an agent responds to any inbound customer message — not just the first one.

  • Starts — Each time a new customer message arrives without a pending agent reply
  • Stops — When the first reply from an agent on your team, or an answer from your Agent Stack, is sent
  • Restarts on reopen? — Only if the clock was still running when the conversation was closed. If an agent reply had already stopped it, it stays stopped after the reopen until the customer’s next message starts it.
  • Best for — Long-running conversations where ongoing responsiveness matters; high-stakes accounts

This is harder to satisfy than First Reply because it fires repeatedly through a conversation. A conversation with three customer follow-ups has three All Messages timers running back-to-back.

Note that this differs from Resolution Time, which restarts on every reopen. After a reopen, the All Messages and Resolution Time rows can therefore read differently: Resolution Time always begins a fresh countdown, while All Messages only restarts if it was still running at close.

Resolution Time

Measures total time from conversation creation to resolution.

  • Starts — When the conversation is created
  • Stops — When the conversation is marked done
  • Restarts on reopen? — Yes — fresh countdown on every reopen, when the conversation comes back into Active
  • Best for — Cases where the outcome matters more than the response cadence — bugs, billing disputes, complex troubleshooting

Resolution targets are typically much longer than reply targets — hours or days versus minutes. A 4-hour First Reply with a 24-hour Resolution is a common shape.

Combining them across policies

Each policy has one metric, with its own priority targets. To cover all three, create three policies, for example:

  • First Reply Time — 8h — 4h — 2h — 30m
  • All Messages — 24h — 8h — 4h — 1h
  • Resolution Time — 5d — 3d — 1d — 4h

A conversation has one SLA policy at a time, so it shows one badge, for that policy’s metric.

Picking which metrics to track

Each row is a starting point, not a single policy. Where a row names more than one metric, create one policy per metric and assign each to different teams or channels, because a conversation is timed by one policy at a time.

  • Standard support team, mixed channels — First Reply on chat channels, Resolution on email
  • High-volume chat where speed is everything — First Reply only (Resolution is hard to bound on chat)
  • Premium support with promised SLAs in contracts — One policy per contracted metric, each assigned to the team or channel the contract covers
  • Internal IT helpdesk with self-set targets — Resolution only (response time is informal)
  • Voice support — First Reply maps poorly to live calls — use Resolution only, or rely on call-handling metrics outside SLA

Don’t track what you can’t influence

Resolution Time on tickets that depend on third-party providers (carriers, banks, manufacturers) is nominally trackable but politically dangerous — agents are held to a target they can’t actually hit. Either exclude those conversations from the SLA via channel/team scoping, or use a much longer Resolution target and rely on Resolution + a manual “external dependency” tag instead.

See also

Tags

Reference