The one on one meeting as a process, weekly meeting limits, and the seven attendee ceiling: Aydin Mirzaee of Fellow

Aydin Mirzaee had never held a one on one meeting in his life until his company was acquired. He did not know what one was or why anyone would do it. That gap is where Fellow came from.

Mirzaee is co-founder and CEO of Fellow, a meeting management platform used across companies including Shopify. Michael uses it at Tucows, which is why this episode exists.

The context Michael sets up front: roughly 55 million meetings happen in the United States each day, and the large majority have no agenda. Shopify's decision to cancel its recurring meetings reportedly reclaimed hundreds of thousands of hours in a fortnight.

Successful in spite of it

Mirzaee's previous company was in the online survey space, bootstrapped to around a hundred people, and sold to SurveyMonkey.

His description of how it ran is the honest version most founders will not give. No one on ones. No quarterly business reviews. No staff meetings. A scrappy startup with everyone everywhere.

Being acquired exposed him to all of that structured management for the first time, through the growth to IPO. And his reflection on his own company is the useful part.

Sometimes you see a successful company doing something and conclude it must be a good thing to do. Sometimes companies are successful in spite of those things. His was the second case. Had they scaled much further they would have been in serious trouble: bursting at the seams, everything duct-taped together, relying on the founders to hold it up.

So the first version of Fellow was a manager's copilot, something to teach the entrepreneur who does not know what they do not know how to run and scale a business.

Then they looked at where managers actually spend their time and found it was roughly half in meetings. Which nobody brags about, and which nobody feels good about, but which is where a great deal of managing actually happens. At some point they decided it was a large enough global problem that if they did not solve it, it might never get solved.

What a one on one meeting is actually for

This is the framing worth stealing, because most people run one on ones without being able to say what they are.

At the highest level it is an alignment meeting: making sure the manager and the direct report agree on the outcomes that need to be achieved.

At another level it is a feedback meeting: how are we doing, how can I help you.

And there are further elements. Career coaching. Support and unblocking.

Once you list them, Mirzaee's point is that these are obviously useful things. Which leads directly to the operator's question: if they are useful, how do you make sure they happen at scale?

The answer has layers. Create a process where every manager does them. Ensure the time exists in the calendar. And then go a level further, by training people or providing a default template, so that the conversation covers the right things rather than whatever comes to mind.

Meetings are processes

He runs the same analysis on the quarterly business review, and it generalizes.

A QBR is fundamentally a review of what happened, comparable to a military after-action review, on a quarterly cadence. But it has other purposes: it lets executives see the rising stars, and it gives everyone a cross-functional understanding of what other teams are working on.

And it is a process rather than an event, because a great deal of preparation feeds into it, and the template improves over time as you learn which prep formats and which questions produce better outcomes.

Which produces the sentence the episode turns on. Recurring meetings are effectively processes put in place to force a series of actions and discussions that are generally beneficial. Retrospectives. Sales discovery calls. Onboarding sessions. All of them.

His analogy for why the calendar is the right instrument is genuinely charming. If you were designing a society, you might decide that gratitude is good and that once a year everyone should be thankful, so you insert a holiday. The calendar ensures certain things happen, on a cadence.

His caution is equally important: do not do them blindly. Running one on ones because you have to is not the same as understanding what they are for.

What templates are actually carrying

Fellow curates expert templates from customers, and Mirzaee's pitch for why explains something about the whole category.

If there is a CEO or COO you admire, would it not be useful to see what their staff meeting looks like?

Because when he asks people about their staff meeting, almost everyone opens by saying they do not know that they do it well, and then reluctantly describes it. Everybody is looking to learn and nobody has anything to compare against.

He is clear that templates must evolve, because what matters is who is in the meeting, what is happening in the business, and whether this is wartime or peacetime.

The place where this is already understood is sales, with discovery meetings, follow-ups, a cadence and a pipeline. His argument is that the same structure belongs in customer success, implementation and onboarding.

The reason organizations buy it

The answer from a Shopify VP, when Mirzaee asked why they liked the product, is the one operators should note.

As a user, they liked it. But what mattered was the confidence that across their organization, everybody was doing it, and doing it the same way.

Because if you stop any leader and ask whether they do one on ones, the answer is that they think so, everyone does them differently, and they do not know how effective they are.

And Mirzaee's case for why this specifically matters: managers are the highest leverage point in an organization. Organizational change happens largely through them. People grow partly because of them. People leave managers rather than companies. And the pivotal interaction is the one on one.

His diagnosis of the blind spot is precise. When we think about operations we think about customer requests arriving and being handled, or orders coming in. We do not think about the standard human interactions happening across the company.

Getting a message to actually land

Mirzaee's example of a hard problem is internal communication, which he says is genuinely difficult from executive level down.

Take any message you want everyone to understand and internalize. It has to be said something like nine times across every channel, and you will still have people asking why nobody told them.

So Fellow built the thing that follows from treating the one on one as infrastructure: an organizational leader can insert a suggested topic into every manager's next one on one agenda, and have the responses bubble back up.

Michael's frustration is the mirror image. Something gets decided, then siloed, then dropped, then forgotten.

Action items, and the accountability half

Mirzaee starts before the action item, at purpose.

Technology changed what a meeting needs to be. It used to be that you brought people together so they could hear something. Which means status update meetings are now largely unnecessary, and his estimate is that they account for around 15% of all meeting content across organizations. He offers it as a straightforward efficiency gain: cut that 15%.

If the purpose is defined, and the purpose is usually to reach an outcome, then the work is detailing how it gets done and what happens next. Fellow's structure is a structured agenda plus a deliberate place for action items, with AI now suggesting them.

The half he thinks people miss is accountability. The recurring complaint is that things get discussed, people agree to do them, and then they do not happen.

His reasoning is worth quoting in spirit: if an action item is not brought up again in subsequent meetings, you are effectively saying it is acceptable for agreed things not to be done. So the design requirement is reminder flows and automatic carry-forward into the next conversation.

And then measurement, from the organizational seat. What percentage of meetings produce action items, and which meetings are the culprits. What percentage are prepared for in advance. What percentage start late. What percentage run over an hour without needing to.

The number that justifies caring: 15 to 20% of an entire payroll is spent on people attending meetings. When you spend that much on anything, you want it done well.

How many hours is too many

Michael's honest answer, when asked what an acceptable number of weekly meeting hours is, is that it depends. Mirzaee agrees that is correct, and then points out that nobody has ever tried to answer it anyway.

His own numbers, derived from Fellow's meeting data by comparing averages against the top decile:

Individual contributors: 7.5 hours a week or less. More than that is too much.

Managers and directors: no more than 15 hours.

Michael's own figure is sometimes 20 to 25 hours, on a week considerably longer than 40, and he suspects the two facts are related.

The important part is enforcement. Most companies would put a number like this in a wiki, which he acknowledges is a reasonable first step and does not work.

Fellow's version builds the guideline into the calendar. When you invite someone who has already passed their threshold, you are told at the moment of inviting. You can still do it, and violating the guideline is made explicit rather than blocked.

His framing of why that is respectful rather than restrictive: it is an educational process.

Participants, not attendees

At a company of a couple of thousand people, Mirzaee says it is not unusual to find a meeting with 63 people in it where only four ever speak.

So the rule is about active participation. If you are not speaking for at least 5% of the meeting, and he acknowledges that is a low bar that many meetings fail, then it is not a meeting you should be in.

The distinction he wants: participants rather than attendees.

And the alternative is not exclusion, it is a better format. Record it, summarize it, send it to the relevant people, let them comment on the recording afterward, and let the original participants read the comments. Which also means people can consume it at double speed, at a time that suits them, possibly while going for a walk.

The seven attendee ceiling

The other threshold is about decisions.

Past seven attendees, Mirzaee says, the ability to reach a decision drops sharply. By around fourteen it is close to zero.

So Fellow prompts at the calendar level. When you go to add the eighth person, it suggests inviting the meeting bot instead, or making them optional.

His broader argument is the one that makes all of this coherent. These norms do not exist. Everyone arrives from a different company carrying different assumptions, and nobody ever states what the norm is here. We define our culture and we define our values. We should define our meeting culture the same way, agree on it, and then encourage and enforce it.

No agenda, no attenda

The controversial one, which he says most companies would not do.

Fellow has a setting that automatically cancels meetings across the company if they do not have an agenda by a specified time, say one hour beforehand.

His promise about the first time you enable it: a lot of people will be angry, because things will disappear from their calendars.

His defense is about behavior change, which he calls the hardest thing that exists. Getting people who have done something one way to do it differently sometimes requires a shock. Miss it once, miss it twice, and by the third time the agenda is there and it is specific.

The motto, which he notes travels well because it gives people the words: no agenda, no attenda.

Michael points out that Darren Murph brought the same phrase to this show, which suggests it is spreading.

The hybrid meeting

Mirzaee is clear that some things should stay synchronous. One on ones, because part of the point is building rapport and knowing people as humans.

But much does not need to be, and his best example comes from Tony Jamous at Oyster.

Before their executive meeting, every executive records a short video update. Mirzaee asked why it had to be video rather than text. The answer: they are remote, and it is not only the bullet points, it is the tone and the facial expression while someone says the thing. Reading text, it is hard to know how to weight something. Hearing it, you know what to pay attention to.

So everyone arrives with the context already absorbed, and the meeting covers only what needs discussing. Fellow does the same for its executive and staff meetings. He calls it a hybrid meeting, and connects it to Amazon's practice of doing substantial work in a memo before the room convenes.

The reason it produces better output is about how people think. Give Mirzaee information and ask him to discuss it immediately and he can, but at a surface level. Give him one night of sleep between receiving it and discussing it, and he arrives with ideas you would not have anticipated.

His fully asynchronous example is an ongoing search issue at Fellow. It started as a weekly synchronous meeting, moved to every two weeks, and is now asynchronous. It stays on the calendar with async marked in the title, so everyone knows they need to act before that end time rather than join anything. Reports go up, people comment, the work moves.

Keeping it on the calendar is deliberate: it programmatically signals that something needs action and attaches a time to it.

Micro-prioritization

The other argument for async, which Mirzaee credits to Dan Martell, is about when your hours are worth more.

Assign your time a large hourly value, whatever number makes you take it seriously. The point is that the value is not evenly distributed. Between nine and noon your brain may be worth ten times that. At six in the evening you are drained and it is worth a fraction.

Asynchronous work lets you match your best tasks to your best hours. Synchronous meetings without hybrid components inject work wherever the calendar had a gap, which is often the wrong hour.

Do not brainstorm and decide in the same meeting

One of the most practical things they have learned.

Reaching a decision uses a different part of your brain than exploring possibilities. Under time pressure, in a single meeting, some people are trying to be creative and others are trying to get back on track.

So be explicit. Run a brainstorming session where you say up front that you are not deciding. Then meet a week later, having considered the options, for a meeting whose purpose is the decision.

The complement is naming roles. There are various frameworks, and what they share is defining who decides, who is informed, and who participates. Explicitly identifying the decision maker for a given meeting makes the discussion more productive on its own.

The most expensive meeting in the company

Asked for his favorite meeting, Mirzaee reasons from cost, which he calculates simply as salaries times time in the room.

By that measure, the most expensive meeting at any company is its town hall or all-staff.

And his observation is that most companies do not spend a proportionate amount of effort making the most of it. It is the opportunity to align the entire company and to get people genuinely excited about what is ahead.

His view of his own job in it: part of being CEO is being chief cheerleader, telling the story and getting people motivated about the future, and the town hall is where that happens at scale.

How they made the company AI-forward

Two mechanisms, both worth copying.

Twice a year, as a remote-first company, Fellow brings everyone together for a three-day summit. Day one is internal talks and customers. Days two and three are an in-person hackathon in which everyone participates, not only developers, because salespeople can work with developers and customer success with marketers, and anyone can build a process.

His method for steering it is the clever part. He deliberately does not suggest ideas, because this is the one occasion where a CEO wants the opposite. The lever he uses instead is the award categories. Make them all AI-based and the outcome takes care of itself.

The second mechanism is the town hall. Every week there is a segment called AI presentations, where anyone can come and show what they have been doing with a tool, and teach everyone else.

His stated reason for the intensity: every company needs to reinvent itself, Fellow included, and if you do not, it is over.

The 5 things I took away from this conversation

1. Say what a one on one is for before you run another one. Alignment, feedback, career coaching, unblocking. I have run these for years without articulating that list, and once it is written down, the question of what belongs on the agenda answers itself.

2. Recurring meetings are processes wearing a calendar invite. This reframe changes what you do with them. You would not run an important business process without a defined output, a template, and periodic improvement. Most recurring meetings have none of the three.

3. Publish the hours, then enforce them where the decision happens. 7.5 hours for individual contributors, 15 for managers. Putting that in a wiki does nothing. Surfacing it at the moment someone sends the invite is the whole difference, and Aydin is right that letting people override it deliberately is what makes it survivable.

4. Seven attendees is the decision ceiling. By fourteen, the group cannot decide anything. That single number resolves a lot of arguments about who needs to be in the room, and it points at the right answer for everyone else: record it and send the summary.

5. Never brainstorm and decide in the same hour. Different modes, in conflict, on a clock. Split them across two meetings a week apart and name the decision maker for the second one.

FAQ

What is a one on one meeting? A recurring conversation between a manager and a direct report. Mirzaee defines it as primarily an alignment meeting, ensuring both agree on the outcomes to be achieved, layered with feedback, career coaching, and unblocking. He argues its value comes from being run consistently and with a defined structure rather than improvised.

What should be on a one on one meeting agenda? The elements that define the meeting's purpose: alignment on outcomes, feedback in both directions, career development, and anything blocking the person's work. Fellow's approach is to supply a default template so managers across an organization cover the same ground rather than each inventing their own version.

How many hours of meetings per week is too many? Fellow's benchmarks, drawn from their meeting data, are 7.5 hours or less for individual contributors and no more than 15 hours for managers and directors. Mirzaee's point is that most organizations have never set a number at all, so nobody knows what excessive looks like.

How many people should be in a meeting? Seven or fewer if a decision needs to be made. Mirzaee says the ability to reach a decision falls sharply above seven attendees and approaches zero around fourteen. His related rule is that anyone not speaking for at least 5% of the meeting should receive a recording and summary instead of an invitation.

What is an asynchronous meeting? A recurring commitment that stays on the calendar but requires no one to join. Participants post updates and comments before the meeting's end time. Mirzaee keeps them on the calendar deliberately, because that is what signals the work needs doing and attaches a deadline to it.

Also mentioned

• Fellow, its meeting guidelines, templates and no agenda no attenda setting
• SurveyMonkey, whose acquisition of Mirzaee's previous company exposed him to structured management
• Shopify, a long-time customer and the company that cancelled its recurring meetings
• Oyster and Tony Jamous, whose pre-recorded video updates precede the exec meeting
• Amazon's six-page memo practice, the same principle applied differently
• Dan Martell's argument that your hourly value is not evenly distributed across the day
• Darren Murph, who brought no agenda, no attenda to this show first

Listen to the full episode

Aydin Mirzaee on Between Two COO's

Between Two COO's is hosted by Michael Koenig. Subscribe on Apple Podcasts, Spotify, or wherever you listen.