Support
Desk insights
What the queue is doing — percentiles rather than averages, business minutes rather than the calendar, and one list that can actually be acted on.
Every ticket carries its own clock. This is the page that adds them up, and the order it puts things in is an argument: what can still be acted on comes first, and the history follows.
Overdue now, and about to be
A breach that already happened is history, not a task
Nobody can un-breach a ticket by working faster today. The counts further down record what happened; the list at the top is only tickets still open and still unanswered, because those are people currently sitting there. Putting settled breaches in a queue is how people learn to scroll past the queue.
*About to be* is a fraction of each ticket's own budget rather than a fixed number of minutes. Twenty minutes left on a four-hour target and twenty minutes left on a three-day one are not the same situation, and a fixed threshold makes every long target useless.
Against what was promised
Measured over tickets raised in the period, not resolved in it — the promise was made when the ticket arrived. Scoring by resolution date would let a bad month be improved by leaving its worst tickets open into the next one.
Tickets with no target are counted separately and never as met
A desk with no SLA policy covering something has not met that target and has not missed it — nothing was promised. Quietly counting those as successes is how an attainment figure gets better the less of the desk is configured, which is the exact opposite of useful. Where nothing at all was promised the rate is blank rather than 0% or 100%.
A clock still running counts in neither column. Treating one as met would make the figure fall every time somebody answered a ticket.
How long it took
Business minutes, and percentiles rather than an average
A ticket raised at five on Friday and answered at nine on Monday took half an hour, not sixty-four. And the distribution is right-skewed: most tickets are answered quickly and a handful sit for weeks, so one of those drags a mean past anything typical. The median answers *usually*; p85 is what a promise should be built on.
The mean is shown anyway, beside the median rather than instead of it — the distance between the two is itself the signal. Far apart means a long tail worth looking into.
Resolved without anybody replying is counted, not excluded
A ticket closed with no answer ever sent has no first-response time, and dropping those would make the first-reply figure improve the more tickets are closed unanswered. It is reported as its own number instead.
Volume
By channel and by priority, with empty rows kept. A channel showing zero this month is a fact worth seeing — it is how somebody notices the chat widget stopped working — and dropping it makes the columns stop adding up to a total anybody can check.
Who is carrying what
This is not a ranking, and should not be used as one
Somebody who takes the hard tickets has a worse median and is doing the harder job. A desk that compared people on this table would get agents picking the easy ones, which is worse for customers than any number on it.
Unassigned work is listed as its own row rather than left out. A desk where nobody is named on a third of the queue has a routing problem, and burying that inside per-person totals is how it goes unnoticed.
What it does not show
- No trend over time. Every figure is one period, and comparing two means opening the page twice. A chart of first-response time by week is the obvious next thing and is not built.
- No cost or margin here. Hours against a ticket are priced when an invoice is raised rather than on this page, and whether a contract already covers them is not answered at all — retainers and included allowances belong to a professional-services module that does not exist.
- Merged duplicates are excluded from every figure. A folded-away ticket is not one that was dealt with, and counting it would inflate both the volume and the resolved count with one conversation twice.