A Meeting Policy That Survives Month Three

A meeting policy that is still working in month three has three properties the ones that die do not. It is written down somewhere a new person finds it without asking. It has an exception process that existed before the first exception. And somebody senior has been visibly inconvenienced by it, in public, at least once.

Everything else is detail. The meeting norms teams argue about for three weeks — the number of no-meeting days, the agenda template, whether thirty minutes is the default or twenty-five — are the parts that turn out not to matter.

Most policies get announced in a channel message, held for about six weeks, and then stop being mentioned. Nobody repeals them. The exceptions accumulate until the rule is the exception, and then everybody privately agrees it did not work. This post is the version that accounts for that: the document, the sequence you roll it out in, and the four failure modes with the countermeasure for each.

Why meeting norms die in month three

Not because they were the wrong norms. Because of how a norm decays.

Week one, the policy is novel and everyone is watching. Week three, the first genuinely urgent thing collides with it, and somebody makes a sensible one-off decision. Week five, there is a second one-off, decided by a different person who does not know about the first. Week eight, a new starter joins and nobody tells them, because the announcement was in a channel they were not in. By month three the rule exists only in the memory of the people who were there when it was announced, and they have stopped enforcing it because enforcing something nobody else remembers makes you the difficult one.

That decay has nothing to do with the content of the policy. It is an artifact of three missing structures:

• No document. The policy lives in a message, which means it has no canonical version, no edit history, and no answer to “what exactly did we agree”. A channel message is unfindable within a month.
• No exception process. So exceptions get granted informally by whoever is asked, which means they are granted every time, because saying no to a specific urgent request with no written rule behind you is socially expensive.
• No enforcement owner. Policies enforced by everybody are enforced by nobody. The person who objects is just a person objecting.

The countermeasures are in the document below. They are the boring parts, and they are the parts that decide whether any of it is true in January.

The policy document

Copy this, cut what does not apply, and put it where your team’s other written decisions live — not in a channel. One page. If it runs to three, you have written a philosophy rather than a policy, and nobody will read it.

Meeting policy — [team name] Owner: [name]. Last reviewed: [date]. Next review: [date + 90 days].

1. Defaults. Meetings are 25 or 50 minutes, not 30 or 60. Any meeting without an agenda in the invite body may be declined without explanation, and declining on that basis is not a performance matter. Recurring meetings expire after 90 days unless the owner renews them.

2. Protected time. [Wednesday] is a no-meeting day for internal meetings. Customer calls, interviews, and incidents are out of scope. Core hours are [10:00–15:00 in the team’s shared window]; meetings outside those hours are opt-in and nobody is marked down for declining one.

3. Who belongs in a room. Every invite names one decision or one output. Attendees are the people needed for it. Everyone else gets the notes. “Optional” is not used — if someone is optional, they are not invited, and they read the notes.

4. Exceptions. Exceptions to section 2 are granted by [the owner named above], asked for in writing, in advance where the situation allows. The reason is recorded in [the same document]. An exception is for one occurrence and does not create a standing one. Incidents and anything with a customer on the other end do not need to ask.

5. Enforcement. [The owner] reviews the exception log monthly and reports the count. Three exceptions in a month for the same recurring meeting means that meeting is reviewed, not that the policy is loosened.

6. What this policy does not do. It does not reduce anyone’s workload, and it will not make a quarter with too much work in it feel different. It moves where the work happens, not how much there is.

Section 6 is not filler. A policy that is sold as a fix for overload gets blamed when people are still overloaded in month two, and that is the most common way a good policy gets repealed for the wrong reason.

Two details worth defending when someone pushes back on them.

Twenty-five and fifty, not thirty and sixty. The ten minutes are not the point. The point is that the number is unusual enough that scheduling it requires a deliberate act, and the default-sized meeting stops being free. A calendar of 25s has gaps in it by construction.

The 90-day expiry on recurring meetings. This is the single highest-value line in the document, because the recurring set is where load accumulates invisibly. It also costs nothing politically: nobody has to cancel anything, they only have to decline to renew, which is a much easier thing to not do. The quarterly review procedure is what that expiry date actually triggers.

Meeting ground rules are not a meeting policy

Worth separating, because searching for meeting norms mostly returns ground rules, and they are not the same object.

Meeting ground rules are behaviours inside the room: start on time, one conversation at a time, phones away, decisions get written down. They are useful, they are cheap, and they are almost entirely uncontroversial — which is why there are so many articles listing them.

A meeting policy is about which meetings exist and who has a claim on your time. That is a question about authority, and it is the reason a policy is hard where ground rules are easy. Ground rules ask people to behave better in a meeting they were always going to attend. A policy tells somebody they cannot book something.

If you only have the appetite for one, take the policy. Ground rules improve meetings that should not be happening. But the two combine well, and the cheapest place to put ground rules is inside the policy document as a short list under section 1, not as a separate initiative with its own announcement. A second announcement halves the attention the first one gets.

The rollout sequence

Order matters more than content here. Most policies are announced before they are agreed, which inverts the only sequence that works.

Week minus two: get the figure. Before you propose anything, count. How many hours the team currently spends in internal meetings, how much of it is recurring, how fragmented the remaining time is. The calendar audit is the procedure, and the cost arithmetic is there if you need the number in money — though be warned that the money version persuades almost nobody and occasionally reads as an accusation.

You need the figure for one reason: it is the before-measurement. Without it you cannot answer “is this working” in month three, and a policy that cannot answer that question is repealed on vibes.

Week minus one: write the document and get one senior yes. Not a consultation, not a survey. One person with enough authority to be the exception-granter, who has read the actual document and agreed to section 4. If you cannot get that, stop. A policy with no named owner is the first failure mode and you already know how it ends.

Ask for it narrowly: “I’d like to try this for a quarter, I’ll own the exception log, and I’ll bring you the numbers in January. Can I put your name on section 4?”

Week zero: announce it with the review date in the first paragraph. The announcement says what changes, when it is reviewed, and that it is reversible. Reversibility is what gets you agreement from the people who think it will not work — you are not asking them to believe in it, you are asking them to let it run for a quarter.

Week one to two: enforce the small stuff visibly. This is the only window in which enforcement is cheap, because the policy is new and nobody has built a habit against it yet. Decline the first agenda-less invite and say which section you are declining under. It will feel petty. It is the single most load-bearing thing you do, because it establishes that the document is a real object rather than an aspiration.

Week six: the check. Covered below, and non-negotiable.

Week thirteen: the review. The date you put in the document. You either renew it, change it, or end it on purpose. Ending on purpose is a legitimate outcome and is vastly better than the drift.

The four ways it dies, and the countermeasure for each

1. Exception creep. The commonest. Exceptions get granted ad hoc by whoever is asked, nobody counts them, and by week ten the no-meeting day has three recurring meetings on it.

Countermeasure: the written exception log in section 4, and the monthly count in section 5. Not to punish anyone — to make the drift visible while it is still small. Three exceptions is not a crisis, it is a signal that one specific meeting needs looking at. Thirty exceptions discovered in January is a dead policy.

2. Leadership not modelling it. The policy holds for everyone except the person who can book over it, and within a month everyone has correctly concluded that it applies to people without power.

Countermeasure: this is the one that needs the public inconvenience. At some point in the first six weeks, the senior person named in section 4 needs to visibly not schedule something, or move something they had scheduled, with a sentence that says why. One instance of that is worth more than the entire document. You cannot manufacture it, but you can ask for it directly when you ask for the senior yes: “the thing that will kill this is if it looks like it doesn’t apply to you.”

3. No written document. The policy lives in a channel message, so there is no canonical version, new people never learn it, and every disagreement about what was agreed is unresolvable.

Countermeasure: the document, where written decisions live, with an owner and a review date on it. Writing it is not the hard part — a policy nobody can find is the same problem as a document nobody read, and the fix is the same: it has to be short, it has to be somewhere findable, and somebody has to reference it in public occasionally or it stops existing.

4. Compression. The meetings do not disappear, they relocate. A no-meeting Wednesday produces a Tuesday with nine meetings on it, and the week is worse than before because the fragmentation moved rather than reduced.

Countermeasure: this is the failure mode that most needs the before-measurement, because it is invisible without one. Total internal meeting hours is the number to watch, not the number of protected days. If hours are flat and only the distribution changed, the policy has not worked yet and the honest thing is to say so at the week-six check rather than in January. Capping hours rather than protecting days is one answer, and it is a different post.

What to do if you cannot set policy

Most readers here cannot. You do not run the team, you have a manager, and announcing a meeting policy is not available to you. The useful version is narrower.

Run the audit anyway and keep it to yourself for now. A figure is the only thing that changes a conversation about meeting load from a complaint into a proposal, and you can produce one without anybody’s permission.

Apply sections 1 and 3 to your own calendar unilaterally. Decline agenda-less invites with a sentence. Stop accepting optional. Put 90-day expiry dates in your own notes for the recurring meetings you own. None of this requires announcing anything, and all of it is defensible one instance at a time if asked. The scripts for doing that, and the costs of each, are the part of the job that actually transfers.

Then propose the smallest version. Not a meeting policy — one rule, for one team, for one quarter, with you owning the log. “Agendas in the invite or it’s fine to decline” is a proposal a manager can say yes to in one message. A six-section document is a proposal that requires a meeting, which is a slightly funny way to begin.

The week-six check

Twenty minutes, in writing, on a date you put in the calendar during week zero. Skipping this is the most reliable predictor of a policy that is dead by January, because week six is the last point at which drift is still small enough to correct without renegotiating the whole thing.

Five questions. Answer them with numbers where you have numbers.

Each bad answer has a specific repair, and all four repairs are cheap in week six and expensive in week thirteen. A missing log starts now. Compression means the next change is a cap on hours rather than another protected day. A new person who has never heard of the rule means the document is not where you think it is. A recurring meeting generating repeat exceptions goes through the fix-or-kill questions — the policy is not the thing that needs changing there, the meeting is.

Write the answers in the document itself, under the review date. Three quarters of that is what makes a renewal decision in month twelve something other than a guess.

Each part of this, on its own

The evidence behind protecting a day and what it does not show, the rollout that does not die, core hours when the team is spread across four time zones, capping hours instead of days, and the autopsy on a policy that died in week six.

• No-Meeting Days: What the Evidence Actually Says — One real study sits behind the no meeting day, and it is weaker than the figures quoted from it. What it found, what it omits, why Zoom ended its own.
• Rolling Out a No-Meeting Day Without It Dying — A no-meeting day dies from exceptions nobody counted. The list you write before launch, the announcement message to send, and the decision you make in week six.
• Core Hours for a Team Across Four Time Zones — A core hours policy across four time zones is mostly arithmetic. Work out the real overlap first, then write the narrow version that survives.
• Meeting Budgets: Capping Hours per Team — A meeting budget caps a team's recurring person-hours, not each person's diary. How to set the number, who has to enforce it, and what breaks first.
• Why Your Meeting Policy Died in Week Six — Why meeting policies fail is diagnosable from the calendar. The four causes, the tell for each, and whether relaunching is worth it or not.