The Recurring Meeting That Quietly Became Something Else
Standing meetings rarely go empty. They go sideways. The weekly slot created for a launch becomes the place escalations get raised, the invite text never changes, and nobody notices because drift only shows up when you compare the meeting to itself six months ago.
The invite still says what it said in March
There is a meeting on your calendar every Thursday at eleven. It was created in March, for a launch, and the title still has the launch codename in it. The launch shipped in May. The meeting is still there and still nearly full.
It is also not a launch meeting any more. Somewhere around June it became the place where anything stuck between two teams gets raised, because those people are already in the room and the hour already exists. Nobody proposed that. Nobody wrote it down. Four people would describe the meeting four different ways if you asked them separately, and none of them would describe the version written on the invite.
This is the ordinary condition of a recurring meeting that is more than about two quarters old. The interesting question is not whether it drifted. It is why nobody in the room can tell you that it did.
The useless meeting is the easy case
Almost every piece of advice about standing meetings targets the empty ones. Cancel the meeting that has no agenda. Kill the meeting where three people talk and eleven listen. That advice is fine and it aims at the cheap problem, because an empty meeting broadcasts its emptiness every single week to everyone sitting in it. Emptiness is self reporting. Eventually somebody has a bad week, looks at the calendar, and deletes it.
The expensive case is the opposite. The meeting is busy. Things get resolved in it. People leave with work. By every test a team would apply, it is a good meeting, so nobody touches it. And the thing it is good at is not the thing the calendar says it is for, which means the attendee list was chosen for a job the meeting stopped doing, the owner is someone whose remit no longer matches the agenda, and anyone who joined the company in July calibrated their understanding of the team from a title that was already wrong.
A meeting that produces nothing wastes an hour. A meeting that produces the wrong kind of value with the wrong people in the room routes decisions past the people who should have made them, every week, on a schedule.
A routine changes just by being performed
Martha Feldman and Brian Pentland spent a 2003 paper in Administrative Science Quarterly arguing against the received view that organizational routines are what makes companies rigid. Their move is to split a routine into two parts. The ostensive aspect is the abstract idea of the routine, the thing people point at when they describe it. The performative aspect is the actual actions, by specific people, at specific times and places, that happen when the routine runs.
The claim that matters here is the direction of travel between them. The idea of the routine guides the performances, which is the obvious half. But the performances also create, maintain and modify the idea. Every time a routine runs slightly differently, the shared sense of what the routine is has quietly moved. Feldman and Pentland describe this as the capability of every routine to generate change simply by continuing to be performed.
A recurring meeting is a routine with a calendar entry attached. So drift is not evidence that a team is undisciplined. It is the default behaviour of the thing. What is unusual about the meeting case is that the ostensive aspect has been written down once, in the invite, in March, and then never updated again, so the gap between the idea and the practice is free to grow without ever showing up anywhere.
Why the standing slot collects other people's problems
Drift has a direction, and there is a fifty year old model that predicts it. Michael Cohen, James March and Johan Olsen published the garbage can model in Administrative Science Quarterly in 1972. Their argument is that organizations are full of problems looking for a venue, solutions looking for a problem to answer, and people looking for something to work on, and that these streams mostly meet by accident. A choice opportunity, in their terms, is any occasion where an organization is expected to produce something that can be called a decision.
Their metaphor is that a choice opportunity works like a bin that participants throw problems and solutions into as those problems and solutions turn up. What ends up in any one bin depends on which other bins are available at that moment, and on the labels attached to them.
A weekly recurring meeting is the most reliably available bin a company has. It arrives on schedule, the right people are already in it, and raising something there costs nobody a new calendar invite. So it absorbs whatever is floating. This also explains a pattern most operators have lived through without naming: a team cuts four standing meetings in a productivity push, and within six weeks the one survivor has become unrecognisable and runs long. The traffic did not disappear. It got routed to the last bin still open.
Nobody says anything, and there is a good reason for that
The obvious objection is that someone in the room should just say the meeting has changed. There is research on exactly that, and it says the two available objections are priced very differently.
Henning Bang and colleagues studied eight top management groups working through 56 agenda items, published in the Scandinavian Journal of Psychology in 2010. Goal clarity and focused communication were both positively related to how effective the team was, and the effect of goal clarity ran partly through focused communication. Then the useful part. Speaking up when the goal of a discussion was unclear improved focused communication, task performance and how people felt about each other. Speaking up when the discussion had gone off track did not improve task performance or member satisfaction, and it was negatively related to relationship quality.
Read that as an operator rather than as a researcher. Asking what we are trying to decide here is a question, it makes the room better, and it costs the asker nothing. Saying this is not what this meeting is for is a verdict on what everyone else has been doing, and the study found it buys no measurable improvement while making the room worse. People are not failing to notice drift. They are correctly reading that naming it in the moment is the expensive option, so the observation stays in their head and the meeting keeps drifting.
Knowing the meeting might be stale does not fix it
The standard remedy is an annual meeting audit, which assumes the problem is that nobody has thought about it. William Samuelson and Richard Zeckhauser tested that assumption, in a general form, in the Journal of Risk and Uncertainty in 1988. Their subjects stuck with whatever option was framed as the current one far more often than a rational model predicts, and two of their findings land directly on a calendar.
The first is that the relative pull of the status quo got stronger as more alternatives were added to the choice set. A calendar with six standing meetings is easier to prune than a calendar with twenty six, and not for the reason people assume. The second is more uncomfortable. They point out that this is not the kind of error that gets fixed by being pointed out. In their debriefs, subjects were readily persuaded that the pattern was real in aggregate, and remained unaware and slightly sceptical that they personally were subject to it. The authors say plainly that they see no obvious escape beyond forcing an even handed comparison of the options.
So an audit that consists of a leader looking at the calendar and asking whether each meeting still feels necessary is running the exact procedure the research says will not work. Feeling necessary is what the status quo does. The only thing that helps is putting the alternatives side by side on evidence, which means you need evidence.
Drift is only visible in the record
Here is the structural problem. Drift is a difference between two points in time. No single instance of the meeting contains it. Week 30 looks like a perfectly reasonable variation on week 29, because it is one, and every week is. The change is real and it is distributed across twenty sessions in increments too small to notice from inside any of them.
That makes the calendar useless as a diagnostic. The calendar holds the title, the time and the attendee list, all of which are the March version. It holds no record of what the meeting actually did in August. Human memory is worse, because the people asked will reconstruct the meeting from its title, which is precisely the artifact that is wrong.
A searchable record of what each session produced is the only instrument that can answer the question, because it is the only place where March and August sit next to each other in comparable form. Pull the last ten instances. Write one line per instance describing what the meeting actually resolved. Do not read the invite until you have the list. Then read the invite. If fewer than half the lines are things a reasonable person would expect under that title, the meeting is not what the calendar says it is, and you now have that on paper rather than as a feeling somebody is reluctant to voice.
Drift is not the failure, mismatch is
The tempting conclusion is that a drifted meeting should be cancelled. Usually it should not. A meeting that drifted toward escalations and is genuinely good at resolving escalations has found a real job. What has failed is the description, and the description is doing work you are not thinking about. It set the attendee list. It set the owner. It tells a new hire what this group is for. It determines who is not invited, which is where the actual damage lives, because the person who should be in an escalation conversation was never invited to a launch meeting.
There are four honest outcomes once you have the list. Rename and reinvite, when the new job is worth doing and the wrong people are in the room. Split, when the list shows two unrelated kinds of work sharing an hour and neither gets a full one. Hand it over, when the work now belongs to a function that is not the current owner's. End it, when the list is mostly status that could be read rather than discussed.
Three of those four keep the meeting. That is worth saying out loud, because a team that believes an audit is a cull will quietly protect every meeting from being examined, and you will get no list at all.
Where Driffle fits, and what it does not solve
Driffle captures meetings without putting a bot in the room and turns what happened into searchable work memory, which is what makes this check cheap enough to actually run. Pulling what a Thursday meeting decided across the last ten weeks is a search rather than an afternoon of scrolling through calendars, chat threads and half remembered docs. Because capture is automatic, the record covers the weeks nobody thought were important, which is exactly where incremental drift hides.
What it does not do is tell you which version of the meeting is the right one. That is a judgement about how the company should be run, and no record makes it for you. It also does not make anyone willing to say the meeting has changed. What it changes is the cost of the claim. Bang and colleagues found that objecting to drift in the room is socially expensive and buys nothing, and part of why is that the objection is one person's impression against everyone else's. A list of ten outcomes taken from the record is not an impression, and it can be sent before the meeting rather than said during it.
It is also worth being clear about what a record cannot repair after the fact. If a decision was made in a drifted meeting without the person who should have made it, having a good record of that decision is better than not having one, and it is not the same as having made it correctly.
A check worth running this quarter
Take the three oldest recurring meetings you own. For each one, pull the last eight instances from the record and write a single line for each about what it actually produced. Keep the invite closed while you do it.
Then open the invite and compare. Look at three things in order: whether the outcomes match the stated purpose, whether the people who owned those outcomes were on the invite, and whether anyone who should have owned one was missing. The third question is the one that produces action, and it is the only one you cannot answer from the calendar alone.
Do this once and you will find at least one meeting that is doing a better job than the one it was hired for, and at least one where the missing person has been missing since June.
Sources
- Reconceptualizing Organizational Routines as a Source of Flexibility and Change (2003), Administrative Science Quarterly 48(1), 94 to 118, on the ostensive and performative aspects of a routine and the finding that performances create, maintain and modify the idea of the routine itself - Martha S. Feldman and Brian T. Pentland, Administrative Science Quarterly
- A Garbage Can Model of Organizational Choice (1972), Administrative Science Quarterly 17(1), 1 to 25, on choice opportunities as occasions where an organization is expected to produce a decision, and on what gets deposited in one depending on which other venues are available and how they are labelled - Michael D. Cohen, James G. March and Johan P. Olsen, Administrative Science Quarterly
- Effectiveness in top management group meetings: the role of goal clarity, focused communication, and learning behavior (2010), Scandinavian Journal of Psychology 51(3), 253 to 261, reporting on eight top management groups across 56 agenda items that speaking up about an unclear goal improved performance and relationship quality while speaking up about an off track discussion did not improve performance and was negatively related to relationship quality - Henning Bang, Synne L. Fuglesang, Mariann R. Ovesen and Dag Erik Eilertsen, Scandinavian Journal of Psychology
- Status Quo Bias in Decision Making (1988), Journal of Risk and Uncertainty 1, 7 to 59, on the finding that the relative pull of the status quo grows with the number of alternatives in the choice set, and that subjects accepted the pattern in aggregate while remaining sceptical it applied to them - William Samuelson and Richard Zeckhauser, Journal of Risk and Uncertainty
FAQ
Is this just an argument for cancelling more meetings?
No, and treating it that way is how the check dies. Three of the four sensible outcomes keep the meeting and change something about it: the name, the invite list, the owner, or the scope. If people believe the exercise is a cull, they will defend every meeting from being examined and you will never see the list you need. Say up front that the goal is to make the description match the work.
How many instances do I need before drift is real?
Eight to ten is enough for a weekly meeting, which is roughly a quarter. Fewer than that and a couple of unusual weeks dominate the sample. Much more than that and you are reading a record of a meeting that has drifted twice, which makes the pattern harder to read rather than clearer. For a monthly meeting, six instances covering half a year works better than trying to hit ten.
What if the meeting drifted and everyone in it is happy?
Then the meeting is probably fine and the description is the thing to fix. Happiness inside the room says nothing about who is outside it. The specific harm from a stale invite is not felt by the attendees, it is felt by the person whose work is now being decided in a room they were never invited to, and by the new joiner who read the title and drew the wrong conclusion about who owns what.
Who should run this, the meeting owner or someone else?
Someone other than the owner should pull the list, and the owner should read it. The owner is the person least able to see the drift, because they have attended every incremental step and their sense of the meeting updated along with it. A chief of staff or an operations lead pulling ten outcome lines and handing them over is doing something the owner structurally cannot do for themselves.
Can we do this from calendar data instead of meeting records?
Calendar data answers a different and smaller question. It shows how many hours a meeting consumes, how many people attend and whether attendance is falling, all of which is useful for load. It cannot show what the meeting produced, and drift is a change in output rather than a change in cost. A meeting can have identical attendance and duration for a year and be doing an entirely different job by the end of it.