Queue history: how to read intake, backlog, wait and SLA per service queue

The Analytics tab of the service console has three different readings, and mixing them up is the fastest way to reach a wrong conclusion. The wallboard answers "how is the queue RIGHT NOW". The service analytics panel answers "how much and how fast in total". Queue history answers the third question

The Analytics tab of the service console has three different readings, and mixing them up is the fastest way to reach a wrong conclusion. The wallboard answers "how is the queue RIGHT NOW". The service analytics panel answers "how much and how fast in total". Queue history answers the third question, which nothing answered before: "did this queue get worse?".

What each column means - Opened: cases that entered the queue on each day of the period. - Resolved: cases that left on each day. They may have been opened earlier, so Opened and Resolved are not two views of the same set. - Open at end: derived, not stored. It is what was open before the period, plus what came in, minus what went out. The product does not keep a history of status changes, so a case that was reopened can make this series drift from the live count on the wallboard. When you need the exact number for right now, the wallboard is the one that rules. - Wait to first reply: from the moment the case was created to the first public reply written by an agent. The customer's own opening message does not count as a reply. - SLA resolution and SLA first reply: the same targets the console charges the agent for, including business hours and pauses. Only cases that actually had a target enter the base. A queue with no target shows a dash, never 0% and never 100%: there was nothing to comply with. - Reopened: share of resolved cases that were reopened at least once. - Handed over: how many times the routing engine took a case from one person and passed it on.

Queues that are not registered The case stores the queue NAME in a text field, so anyone can type a name that was never registered. Those appear here marked "not registered", with the name exactly as it was typed. They are not hidden and they are not lumped into "other": a queue nobody registered cannot be routed, cannot have a cap and has no working hours. Register it in Settings > Queues and it starts being managed like the others.

Case sensitivity, accents and extra spaces do not split a queue: "Support", "support" and " Support " are the same line.

Comparison with the previous period Each row carries the change against the immediately preceding period of the same length. Green means better, and for wait time better means going DOWN, which is why the arrow direction and the color do not always agree.

Where the numbers come from The history is built once a day by the system and read in milliseconds after that. Counting happens in the database, with no ceiling: a day with tens of thousands of cases enters whole. If some day of the period was never built, the screen says so instead of showing a gap that looks like a quiet day. Use Generate history to fill in past days or to refresh after a correction.

Export and dashboards Export CSV downloads exactly the table on screen, with one line per queue. Build my dashboard adds ready-made cards for a role (agent, supervisor or manager) to your personal dashboard. They become normal charts from that moment on: rename, edit or delete them freely.

What this reading does not answer There is no cost per case: the product has no internal cost per case to divide. Handle time per agent is an ESTIMATE based on activity in the conversation, not a stopwatch, and it is labeled as such wherever it appears.

Open this article inside the system →