Fast support matters, but one number cannot describe the customer’s entire experience. An instant first reply followed by an hour of silence is not a fast resolution, while a long conversation does not always mean weak performance—it may require a payment check, technical investigation, or an answer from the customer. Lavenity therefore separates speed into several measures instead of presenting one universal timer.
Separate waiting into four different measures
Median first-response time shows how long a typical new conversation waits for the first reaction. Median response time describes the usual gap between messages in an active exchange. Time to close covers the path from the start of a conversation to its completion, while handling time focuses on a teammate’s active work. Together, they separate team availability from conversational pace and the complexity of resolution.
Reports use medians because a median is the middle of an ordered set: half the conversations are faster and half are slower. One thread accidentally left open for a week would pull an average sharply upward, but it will not distort the median to the same extent. A median still hides the worst cases, so read it with Open and Unassigned conversations rather than treating it as proof that nobody waits too long.
Compare like-for-like context
Compare the same channels over matching periods. Live chat usually carries a faster expectation than an asynchronous messenger, and an overnight shift has different coverage from business hours. The overall median may improve simply because the team received more straightforward website chats this week. A channel filter separates an actual process improvement from a change in the composition of incoming work.
Read the four measures as a sequence. A long first response followed by a normal conversational pace points to coverage or assignment. A quick start followed by slow replies may indicate that an operator is carrying too many active threads. A normal exchange with a long time to close often indicates an unclear completion rule, an external dependency, or conversations the team forgets to close after resolving the issue.
Set targets that protect quality
Set a target from your own baseline and customer promise, not from a random industry number. First collect several stable periods, define the hours when the team is truly available, and find where delays repeat. Then choose one operational change: a rotation for Unassigned, a saved reply for a frequent question, a clearer handoff, or a closing rule. The target should describe the intended experience, such as a predictable first reaction during business hours, rather than merely “lower the median.”
Do not reward speed without progress. If the team sends a brief “we received your message” only to satisfy the timer, first response looks better while the customer remains stuck. Check speed against the handling flow, conversations with no reply, Open and Closed states, and the reasons customers return. A healthy metric locates delay; it should never encourage people to optimize a number at the expense of resolution quality.
Investigate one slow slice
Choose a range where the median became meaningfully slower and narrow the data one channel at a time. For the affected channel, compare first-response time with subsequent response and closing times. Then inspect volume, Unassigned, and the teammate distribution. This sequence separates three causes: requests wait too long before assignment, operators move slowly through active exchanges, or completed conversations stay open because the process has no clear closing rule.
After a change, do not judge the result the next day when normal volume is too low. Wait for a comparable complete period and inspect neighboring signals as well as the median: did conversations with no reply increase, did the team begin closing threads prematurely, or did work concentrate on one teammate? A real improvement preserves answer quality and works across the queue instead of moving the problem into another metric.
Frequently asked question
Why does the report show a median instead of an average time?
A median better describes the typical conversation because one exceptionally long or forgotten thread cannot pull the entire number sharply upward. It does not reveal the worst waits, however, so read it with Open and Unassigned conversations and investigate repeated delays within a specific channel.