A meeting audit should be treated as an operating review, not a calendar-cleaning exercise. Its goal is to recover decision time and attention, even if the final meeting count barely changes. A ten-person team’s weekly one-hour meeting consumes 520 person-hours a year before preparation; cutting it to 45 minutes returns 130 hours. That arithmetic matters, but it can mislead: a 20-minute huddle that prevents a week of duplicated work may be more valuable than a two-hour meeting everyone dislikes.
That is the strongest case against aggressive meeting cuts. Meetings carry context that is hard to document, give new employees access to informal knowledge, and create a shared place to resolve disagreement. For distributed teams, live time can protect people from decisions being made by whoever happens to be online first. Remove every recurring meeting and the work does not vanish. It moves into messages, documents, one-to-one conversations, and duplicated effort.
The fair response is not to defend every invitation. It is to audit the meeting series itself: its purpose, decision rights, participants, cadence, and actual output. In The Surprising Science of Meetings, meeting researcher Steven Rogelberg makes a similar case for improving meeting design rather than chasing a universal ban. There is no reliable number of weekly meeting hours that works for every team. There is a practical method for finding which meetings earn their cost.
Audit the series, not the calendar slot
Most calendar cleanups inspect individual invitations. That is the wrong unit. Audit the recurring series: the weekly staff call, monthly operating review, daily stand-up, or biweekly cross-functional sync. A single canceled event feels productive for an afternoon. A redesigned series changes the system that created the load.
Start with two ordinary workweeks, not a holiday week, product launch, or annual planning period. For monthly or quarterly meetings, add the last two occurrences so you can see whether their stated purpose actually appears. Use one row per series, plus any one-off meeting that involved at least five people or lasted more than an hour. Record the owner, stated purpose, cadence, duration, required attendees, invited versus actual attendance, preparation, follow-up, recent decisions, and the last two outputs. Read the notes or agendas where they exist. “Weekly sync” is a calendar label, not a job description.
Calculate a rough person-hour cost using required attendees rather than everyone on the invite. Eight people in a 45-minute meeting consume six person-hours. Multiply that by the number of occurrences, then add a separate estimate for preparation and follow-up. The number is not a finance-grade valuation; it is a way to make recurring cost visible. It also stops a meeting owner from claiming a large saving by removing optional names who never attended in the first place.
Do not begin with a team poll asking people to rate every meeting from one to five. Satisfaction is useful context, but it is a poor verdict. An incident review may be tense and still prevent a repeat failure. A comfortable status meeting may leave no decision, owner, or changed plan. First inspect what the meeting produces; use sentiment to explain the result.
For most teams, a light scan each quarter is enough. Run a fuller meeting audit after a reorganization, a change in leadership, or a major shift in the team’s work. The point is to examine a changed operating system, not to create another recurring administrative ritual.
Give every recurring meeting an observable job
Every meeting should be able to complete this sentence: “By the end, participants will…” The ending might be “choose between two vendors,” “assign owners to the top three risks,” “practice a difficult conversation,” or “leave with one clear priority.” “Create alignment” is a starting intention. It needs to become an observable result.
Use a cancellation test. If the next two instances disappeared, what decision, handoff, or relationship outcome would be at risk? “People would feel out of the loop” is a signal to define the information they need and the point at which they need it. “The launch decision would slip by a week” is a more useful answer because it identifies a real consequence.
Then clarify decision rights. A lightweight DACI model can help: one driver, one approver, named contributors, and people who need to be informed. If a meeting has twelve contributors and no approver, it may be collecting opinions without creating a decision. If everybody is marked as informed but nobody owns the follow-up, the meeting is likely exporting work into ambiguity.
Attendance deserves its own inspection. The person who must decide may need to attend. A subject-matter expert may only need to provide input before the meeting. A senior leader may need the outcome afterward rather than a seat at the table. Presence is not proof of relevance, and removing someone from a meeting without giving them a reliable way to contribute is not a saving; it is exclusion with better arithmetic.
Use the same discipline for meetings whose output is less formal. A one-to-one may produce clarity about a blocker, a development commitment, or an early warning that should not become a document. A retrospective may produce one experiment for the next sprint. A team connection session may create belonging rather than a decision. These meetings are not exempt from review, but they need the right evidence. A social gathering should not be forced to pretend it has a project plan, and a coaching conversation should not be judged by the number of action items it generates.
At the end of the review, give the series a specific decision: keep it, shorten it, reduce its frequency, rotate attendance, redesign it, or end it. Assign an owner and a review date. “We should make it better” is not an audit result.
Replace what you remove
The counterintuitive part of a good meeting audit is that it may create work before it reduces work. A team may need a clearer decision note, a better pre-read, or an owner for follow-up before a large recurring meeting can shrink. That is a worthwhile exchange if the new artifact replaces repeated explanation rather than adding another layer of reporting.
For example, a weekly project-status meeting can become a written update by Tuesday noon plus a 20-minute exception review for issues above an agreed threshold. A broad decision meeting can become a short pre-read, a 48-hour comment window, and a 20-minute session attended by the driver, approver, and essential contributors. The replacement should make the route to a decision clearer, not merely move the same crowd into a document.
Asynchronous work has a cost. It favors people with more time, stronger written fluency, and better access to the relevant information. It can also turn one meeting into ten notifications and a long chain of clarifying messages. If the replacement requires everyone to read every update and answer every thread, the hours have moved rather than disappeared. Some disagreements, sensitive conversations, and accessibility needs are better served by live discussion.
Write a short meeting-change note for each redesigned series: its new purpose, required attendees, expected output, replacement process, and review date. For a weekly meeting, test the change over the next two occurrences. For a monthly meeting, test it twice before declaring success. State the failure condition in advance: a missed regulatory deadline, an unresolved customer escalation, a decision that takes longer than the agreed window, or a measurable increase in rework.
A protected half-day for focused work can be useful, but only if leaders respect it and urgent exceptions have a clear rule. A company-wide “no meeting day” often pushes meetings into the remaining days and makes the calendar more brittle. Local experiments are less impressive and more likely to survive.
Measure whether the meeting culture improved
Fewer meetings are a means, not the outcome. Track a small baseline for two weeks and check it again after the pilot. Useful measures include:
- recurring person-hours by team and role;
- time from an issue being raised to a decision being recorded;
- actions carried over from one meeting to the next;
- rework caused by unclear or delayed decisions;
- the number of uninterrupted work blocks of at least 60 minutes; and
- a two-question pulse: “I know which meetings require my presence” and “Important decisions happen within a reasonable time.”
The Kirkpatrick model’s distinction between reaction, behavior, and results is useful here. A meeting receiving positive ratings is reaction data. A decision being made, communicated, and carried out is behavior and result data. Keep both, but do not confuse a pleasant hour with effective coordination.
There are real limits to this method. The evidence for a universal productivity gain from reducing meeting volume is thin because work is transferred between calendars, messages, documents, and informal conversations. This fails when executives audit everyone else’s meetings but keep their own recurring forums untouched. It also fails when unclear priorities or decision rights are the real cause of the meeting load. Cutting invitations will not repair either problem.
Incident response, customer escalation, safety reviews, onboarding, and some one-to-ones require live attention. A new hire who cannot find help, or an overnight support worker who is pushed into unpaid message triage, is not evidence of a successful audit. Before canceling a meeting, ask who loses access to information, coaching, or decision-making when it goes away.
On Friday, choose one recurring series with a visible person-hour cost. Bring its last two agendas, write its expected output in one sentence, assign the decision roles, and either cancel the next occurrence or run a shorter version with only the required people. Review what broke after 14 days. If nothing important broke, you have evidence for changing the series—and a more credible foundation for changing your meeting culture than another complaint about too many meetings.