Skip to main content

​
Read first-reply-time metrics: what the number includes

First-reply time (also called first-response time) is the single most-referenced SLA number in the inbox, and the most misread. This page is the semantic: what the metric actually measures, where the 30-day figures live, and what does and does not count toward it. Once you read it correctly, the SLA timers engine and the SLA report pages take over the mechanics and the live breach view.

​
1. First reply time vs resolution time

Two clocks measure different promises:
  • First reply time — elapsed time from the customer’s first inbound message to the first reply your team sends. It measures how long a customer sat with no answer at all.
  • Resolution time — elapsed time from the conversation opening to the moment it closes. It measures how long the whole issue took.
A five-minute resolution on an old thread can still be a first-reply breach; a one-minute first reply followed by a three-day resolution is a resolution breach. Read them separately, always. (Third clock, for completeness: next-response time — the wait after the customer’s follow-up message — restarts each re-contact.) One predicate decides whether the first-reply clock exists at all: it starts only on an inbound message. An agent-initiated thread the customer never answered — a campaign, a proactive note, an outbound call — carries no first-reply clock, so it can never count as a first-reply breach and never dilutes the metric. This matters in the denominators below.

​
2. Where the 30-day first-reply metric is read

Three surfaces, three different questions:
  • Inbox → Analytics → Team performance — the review surface. The First-response SLA card shows the share of first replies inside target as a percentage (green above 95%, amber 80–95%, red below 80%); the Per-agent breakdown table carries the same percentage per agent, and expanding More metrics adds Avg first reply — the raw seconds behind the percentage. Pick Last 30 days in the window selector and the row you screenshot in the weekly review is the standard 30-day first-reply figure.
  • Inbox → SLA report — the live surface. Answers “how many threads are over the first-reply line right now,” plus a per-channel breakdown. It always shows the current state, not the historical average. See Read the inbox SLA report.
  • SLA escalation policies — not a metric reader, but the place that consumes the same breach logic at queue level: a per-digital-queue escalation policy decides what happens when first-reply breaches keep landing. Read the metric in Analytics or the SLA report; act on it with escalation policies.
The discrepancy to expect: the Analytics percentage is a 30-day attainment figure; the SLA report count is a live breach snapshot. They will not match, and both are right.

​
3. How channel mix widens or narrows the window

The first-reply average is a mix, and the mix moves when channel volume moves. If your WhatsApp policy runs a 30-minute target and email runs a four-hour target, a week of heavy WhatsApp volume pulls the aggregate average down even if per-channel speed is unchanged — and a late-night-hours SMS surge pulls it up when business_hours_only policies overnight the clock. Practical rule: before calling a trend real, open the SLA by channel row on the SLA report and compare per-channel attainment, and re-read the aggregate only after channel mix is stable. A drill into the Per-agent table’s More metrics resolves whether the raw average moved for one channel or for all of them.

​
4. How dispositions affect first-reply accounting

Dispositions are conversation-level outcome labels (Resolved, Refund issued, Escalated). They attach to the close of a conversation — that is, to resolution, not to first reply. First-reply time starts at the first inbound message and stops at the first outbound reply, regardless of how the conversation is later dispositioned. Where dispositions do enter the first-reply picture:
  • The label is stamped at close, so a conversation closed without a first reply ever sent (spam, a wrong-number thread) still counts toward resolution time but leaves the first-reply clock unanswered. If you disposition these into their own label (for example, “No response needed”), filter your reading to exclude them — otherwise they inflate the resolution average while the first-response percentage stays clean.
  • Re-opened conversations (a reopen within your re-contact window) start the resolution clock again, but the first-reply clock was already answered once and stays answered. Cleanup dispositions keep these threads out of the reopened set.
For designing those label sets so reporting stays clean, see Design inbox dispositions.

​
5. Example calculations

Worked from the case style used in Unified inbox SLA engine. Synthetic default thresholds apply until you attach a policy (4h first response, 24h resolution, business hours only): Example A — first reply inside target, resolution over. A customer writes at 09:12; the operator answers at 09:37 (25-minute first reply), and the conversation closes at 11:47 (2h35m resolution). Result: first response not breached (under 4h), resolution not breached (under 24h) — but read the clocks apart, because a longer resolution would have been a resolution breach while the first reply stayed clean. Example B — agent-initiated thread, no first-reply clock. An outbound campaign message goes out at 14:00; the customer never replies. Result: no first-reply clock exists, so no first-response breach can ever fire; resolution time runs from open to close against the 24h default (or your policy target). Do not let such threads dilute your first-response percentage — they are excluded by the inbound-message predicate, not by you. Example C — business-hours policy freezes the first-reply clock. A customer writes Friday 17:40; your workspace posts business hours of Mon–Fri 09:00–18:00 and the policy has business_hours_only: true. The operator replies Monday 09:05. Result: first reply counts only the business hours between the message and the answer — 17:40–18:00 (20 minutes) plus 5 minutes Monday morning = 25 minutes against a 30-minute target: inside. Example D — 30-day attainment percentage from raw data. On the Team performance page at Last 30 days: 1,842 inbound conversations with a first reply; 1,774 sent inside target. First-response SLA = 1,774 / 1,842 = 96.3% — green. The Avg first reply column reads the raw seconds of those 1,842 first responses; the percentage and the average answer different questions and both appear on the page.

​
See also