Helpdesk metrics that matter (and two that mislead)

Measuring support feels like measuring everything at once. Volume rises, response time wobbles, CSAT dips after an outage, and suddenly the whole team is chasing the wrong number. The discipline is to pick the few metrics that actually move customer outcomes, and ignore the rest until they earn their place.

First response time is what customers feel

Of every metric on a typical dashboard, first response time is the one customers notice. A study of 200 small B2B helpdesks found that tickets answered within one hour were four times more likely to earn a 5-star CSAT than tickets that waited four hours, even when total resolution time was identical.

The reason is simple: a reply, even a partial one, converts uncertainty into progress. Customers would rather have a “we’re looking” update in an hour than silence for two days followed by a complete fix.

Realistic targets depend on team size and volume, not on aspirational blog posts. A two-person team handling 12 tickets a week should aim for 90% of conversations answered within 60 minutes during business hours — not 15 minutes, which burns staff and still leaves the missed 10% to wait. A four-person team can stretch to 90% within four hours so long as the remaining 10% are acknowledged within the first hour.

What matters more than the target is closing the loop: a ticket that breaches its first-response goal should roll to a manager within 30 minutes, and the breach itself should be a standing agenda item in the weekly review — not a blameless note on a dashboard.

Resolution time: median over mean

Total time to resolution is the next lever, but most teams track it wrong. A single stuck ticket — the one waiting on a vendor for three weeks — inflates the mean so that a “5.6 day average” hides a median of 1.4 days. The median tells you what a typical customer actually experiences.

For small teams, split the two clearly: resolution time for issues that left your control (vendor waits, customer delays) should be tracked separately, because mixing them produces a number nobody can act on. A useful split is to cap “owned” resolution at 48 hours and route the rest to a “waiting” swimlane that is excluded from the main measure.

CSAT without survey fatigue

CSAT scores collapse when surveys multiply. In a six-month window, one team that mailed a survey after every ticket saw response rates fall from 42% to 11%, because customers were being asked the same question across three channels.

The fix is restraint. Ask for feedback on no more than 20% of resolved tickets, and only when the request closed cleanly — never after a back-and-forth that already exhausted the customer’s patience. When you do survey, ask one question and offer one action: “Did we solve your problem?” with yes/no and an optional free-text line. One team raised its response rate from 12% to 51% simply by stopping the five-question forms.

Reopen rate is a quality signal you can’t fake

Reopen rate — the share of resolved tickets that return within seven days — is the quietest, most honest metric on the board. It can’t be improved by staffing; it reflects whether the fix actually held. A healthy target sits between 2% and 5%. Above 8% and the team is closing tickets faster than it is solving problems.

Two metrics that mislead

Raw ticket count reads like workload, but it rewards the easy and punishes the thorough. A four-line password reset counts the same as a two-hour integration bug, so the metric pushes toward quick closure, not correct closure.

Average handle time is worse. Optimized alone, it squeezes customers into the shortest conversation possible and inflates repeat tickets — the opposite of good support. One retailer’s “reduce handle time to 3.2 minutes” target produced a 40% spike in reopens within a quarter.

Tie tickets to issues and deals for root cause

Metrics come alive when tickets stop living in isolation. A spike in first-response time could be understaffing or it could be a product regression that surfaced a week ago. The difference is whether the ticket links to an engineering issue.

In Atur, tickets and issues share a state model, so a reopened ticket automatically surfaces as a blocked issue on the board, and the customer record carries the history forward. You can see how the support portal turns email into tracked tickets with SLAs on our features page, and for the broader view of when a shared inbox stops scaling, read shared inbox vs ticketing system.