Key takeaways
- An average call duration almost never describes a real call. A team that clears half its calls in under thirty seconds and spends twenty minutes on the other half shows the same average as a perfectly steady team.
- A new report shows the full distribution, team by team: one bar per team, split into six duration bands, each block carrying its share and its number of conversations.
- The bands are not fixed. They are computed from the conversations in your scope, so that the six blocks carry comparable volumes instead of following an arbitrary grid.
- Clicking a block opens the matching conversations, filtered on the team and on the band: from statistics to listening in one move.
- The report counts every audio conversation, including those not yet analysed: duration is known from the moment a call is imported.
What the average leaves out
Ask for a team's average call duration and you will get a number. Three minutes and twelve seconds, say. That number has the merit of existing, and the drawback of describing no particular call.
Take two teams with the same average. The first handles steady calls, nearly all between two and four minutes. The second clears half its calls in under thirty seconds and spends twenty minutes on the other half. Same average, two different jobs, two different problems, and two management decisions that have nothing in common.
This is not a textbook case. Very short calls usually mean something: an immediate transfer, a routing mistake, a customer hanging up before being taken care of, a misdialled number. Very long calls mean something else: a complex case, an advisor hunting for information, a procedure that gets in the way. The average melts both into a figure that belongs to no one.
One bar per team, six duration bands
The new Call duration by team report shows the distribution rather than the summary. Each team occupies one row, as a bar normalised to 100% and split into six duration bands. Each block displays the share it represents and the matching number of conversations.
Rows read independently of one another: two teams with very different volumes stay comparable, since every bar is normalised. A team handling four thousand calls and one handling three hundred are compared at a glance on the shape of their bar, not its length.
Teams appear in alphabetical order of their full path, the very one shown by the team selector in the filters. A team therefore sits in the same place in the filter and in the chart, and two teams with the same name under different branches never get confused.
Bands computed from your own data
The first instinct when building this kind of report is to fix the bands: under thirty seconds, one to three minutes, three to ten minutes, and so on. It is simple, and it is often unusable. On an appointment-setting activity where nearly every call runs under two minutes, a grid like that files 90% of the volume into one box and leaves the others empty. Nothing shows.
The bands in this report are therefore derived from your conversations. The computation looks for the six intervals that best spread the volume of your scope, picking the boundaries from a scale of readable thresholds: fifteen seconds, thirty seconds, one minute, two minutes, five minutes, and so on up to thirty minutes. On a short-call activity the boundaries come down and the report tells fifteen seconds apart from forty-five. On a technical support activity they climb.
Two practical consequences. The boundaries change when you change scope or period: that is the price of bands that carry information. And when the volume is too small or too concentrated for a split to mean anything, the report falls back on a default grid rather than producing empty bands.
From the chart to the conversations
A report that hands you a figure without letting you open it leaves the reader with an unanswered question: why.
Every block is clickable. One click opens, in a new tab, the list of conversations it represents, filtered on the team of the row and on the boundaries of the band, keeping the filters of the report. The chart stays visible behind it: you can open several blocks and compare them.
This is where the report earns its keep. Noticing that 12% of a team's calls run under fifteen seconds says nothing on its own. Listening to ten of them says everything: they are hung up before pickup, or transferred for lack of clearance, or concentrated on one specific time slot. The statistic points at where to look, listening provides the answer.
Start with the extremes. Open the shortest band of the team that has the most of them, then the longest band. Those are the two populations where actionable anomalies hide: the calls that should never have landed there, and the ones that should have been simpler to handle. The middle of the distribution mostly confirms what you already know.
A deliberately wide scope
This report counts every audio conversation in your scope, including those that have not been analysed yet. That is a deliberate difference with the score-based reports, and a banner at the top of the chart says so.
The reason is simple: a call's duration is known as soon as it is imported, well before analysis completes. A duration report showing only analysed conversations would describe the progress of processing rather than the phone activity of the teams. The volumes shown here will therefore often exceed those of your quality reports, and that is intended.
Written conversations are left out: an email or a chat has no duration to spread.
Filters, export and sharing
The report relies on the same filters as the others: period, teams, advisors, scorecards, criteria, attributes. The duration filter stays available, with a warning displayed while it is active, since it mechanically truncates the very distribution you are looking at.
Like every report, it is shared by URL, added to favourites, exported block by block as PNG and as a PDF with its cover page and its scope. The PDF carries exactly the same reading as the screen, bands and volumes included, so it can travel to a management meeting without a screenshot.
What this report does not measure
One useful point before opening it in front of a team: duration is not a quality indicator.
A short call can be excellent, a long call can be exemplary. Nothing in this report supports saying that one team works better than another, and using it as a ranking would be a misreading. What it shows are activity profiles and their anomalies: an unusually crowded band, an unexpected gap between two teams supposedly doing the same job, a shift from one month to the next.
That is also why it deliberately stops at team level. The question it helps to ask is "why does this team have this profile", not "who makes the longest calls". To assess the quality of a conversation, your scorecards remain the tool, and the duration report serves to choose which ones to listen to.
Want to see this report on your own conversations? Our teams will walk you through it on your data.