A weekly metrics meeting earns its hour when it ends with one decision and one owner, and fails when it ends with everyone having seen the numbers. Four charts are enough to get there: ticket volume, backlog age, response time and resolution time. Anything beyond those four is a different meeting.
I ran this review weekly at a UK platform business in the late 2000s. We printed the pack the afternoon before and stuck the charts to a whiteboard, and the useful part was always the marker pen: whatever got written under a chart was what the week was about. The version below is the cleaned-up agenda, with the Excel behind it.
What does a one-hour metrics meeting look like?
The hour is split into five blocks, and the times are hard limits. This is an illustrative agenda, so adjust the minutes to your team, but keep the shape.
| Minutes | Block | Who leads |
|---|---|---|
| 0-5 | Last week’s decision: did it happen? | Meeting owner |
| 5-25 | The four charts, about five minutes each | A different shift lead per chart |
| 25-45 | The one question the charts raise | Whoever owns that queue |
| 45-55 | The decision, written on the board | Meeting owner |
| 55-60 | Actions, owners, dates | Note-taker |
Rotating the presenter does more than share the load. Call Centre Helper’s advice on rotating who presents the performance data, with the manager adding commentary afterwards, matches what I saw: a shift lead who has to explain a chart in front of peers checks it first, and bad data gets caught before the meeting rather than during it.
The first five minutes matter most and get skipped most. If last week’s decision was “move two people to the evening shift” and nobody can say whether it happened, the rest of the hour is theatre.
Which four charts earn a place on the wall?
Volume tells you what came in. Backlog age tells you what you haven’t dealt with. Response time tells you how long people waited to hear from you, and resolution time tells you how long until it was fixed. Together they show a queue from both ends, which is why I stopped adding a fifth.
The one most teams get wrong is backlog age. A backlog count of 10 tickets sounds fine, but it hides the difference between ten tickets raised this morning and ten raised three weeks ago. Age is what your customers feel, and it’s the chart that usually produces the week’s decision. The ticket categories you pick decide whether you can split these charts by type of work, so settle that taxonomy first or every chart will argue with the last.
Keep response and resolution on the same page, because they pull against each other. Pick the response measure deliberately too: average speed of answer and service level tell different stories, and the chart should show only one of them. Pushing a fast first reply is easy; an acknowledgement that says nothing will do it. If response time improves while resolution time slips, you’ve bought a number, not a service.
How do you build the backlog-age chart in Excel?
Keep one sheet of tickets, exported each Monday: ticket ID in column A, date opened in B, date closed in C (blank while open). Then:
- D2, age in days for open tickets only:
=IF(C2="",TODAY()-B2,""). Fill down the whole export. - Age buckets in a small table:
=COUNTIFS($D$2:$D$500,">=0",$D$2:$D$500,"<=2")for 0-2 days, then the same with 3 and 7, 8 and 14, and a single">=15"for the oldest. COUNTIFS arrived in Excel 2007; on 2003 use SUMPRODUCT instead. - Median age:
=MEDIAN($D$2:$D$500). MEDIAN skips the empty text cells, so closed tickets drop out on their own. - Average age:
=AVERAGE($D$2:$D$500), shown beside the median on purpose.
Here is a small example, illustrative only. Ten open tickets are aged 0, 1, 1, 2, 3, 5, 6, 9, 14 and 21 days. The buckets come out at 4 tickets in 0-2 days, 3 in 3-7, 2 in 8-14 and 1 over 14. The average is 62 divided by 10, which is 6.2 days, but the median is 4.0. That gap is the signal: one 21-day ticket drags the average up while the typical ticket is much younger. Show both, and ask about the oldest one by name.
What gets dropped when the meeting overruns?
Decide before the meeting starts, not during it. When the clock hits 25 minutes and the charts aren’t done, cut in this order: the resolution chart discussion first (read the number, skip the debate), then response time, then volume. Backlog age and the decision block are never cut. If a chart eats its five minutes and there is still a question, park the question in a log and give it to the owner.
Two other rules save more time than any agenda. Nobody gets a chart they didn’t ask for, which means a new chart has to replace an old one. And anything that needs more than two minutes of discussion leaves the room as an action for a smaller group. A meeting of eight people debating one queue’s workflow costs eight hours of the week.
How do you know the meeting is working?
Count decisions, not attendance. After eight weeks, look at the board photos: if at least six show a decision with a name and a date next to it, the format is paying for itself. If they show charts and no writing, shrink the meeting to thirty minutes and make the decision block mandatory. The test I’d keep is simple. Take the oldest ticket in the backlog and ask who in the room would be surprised to hear its age. If the answer is nobody, the pack is doing its job.
Hutch Morzaria is a Director-level CX and Support Leadership professional with 19 years of experience building global support organizations across SaaS, Fintech, and enterprise technology. He has hired dozens of support and CX leaders across his career and holds ITIL Expert certification across V3 and V4.
