Patrick McKendry

Shopify made headlines this week by welcoming employees back from the holidays with one last gift - a meeting purge. After the purge, all recurring meetings of more than 2 people were canceled, Wednesday became blocked as a “no meeting day”, and any meetings of more than 50 people were permissible only on Thursday between 11am and 5pm¹. CEO and co-founder Tobi Lutke quoted, “The best thing founders can do is subtraction.”² On one hand, meeting culls are popular and this seems like yet another trite effort by yet another C-suite to drive culture with broad brush rule implementation. The move carries additional significance in light of a recent round of lay-offs³ at Shopify. Reduced meeting counts surely free up individuals’s task capacity to pick up any workload redistribution that comes with team down-sizing. It is also an interesting move given Shopify’s commitment to remote and hybrid work⁴.For many remote employees, organized meetings are one of the few - if only - opportunities to experience and participate in the team culture.

On the other hand, everyone hates meetings. Studies show that employees feel 50% of all meetings are a waste of time⁵with about $37 billion lost to meetings annually⁶ and 71% of senior managers describing meetings as “unproductive and inefficient”⁷. This pain has only increased as a result of the COVID-19 pandemic. According to Microsoft, as companies moved to remote and hybrid set-ups, the average Teams product user experienced a 153% increase in meetings scheduled. With such a meeting load, the end result is what you expect. In the same study, Microsoft found that 42% of meeting participants multi-tasked, sending emails or other messages during their meetings⁸. This risks spiraling out of control as lower meeting engagement leads to less productive meetings, which leads to more meeting scheduling in order to arrive at decisions, which leads to more multitasking in meetings, which leads to lower engagement, and onwards and downwards.

More specifically, what everyone obviously really hates is bad meetings. Yet, broad-reaching rules on meeting allowance limits all meetings - both good and bad. These rules are too simplistic. Similarly at fault are simplistic rules-of-thumb for what it takes to make a good meeting. Of course, if you do not send a meeting agenda in advance or conclude a session with takeaways START DOING THAT IMMEDIATELY. But this is still not enough to make a meeting good. There is already a tremendous amount of content written around improving meetings - read from leading expert Dr. Steven Rogelberg if you want to get up to speed quickly⁹.

Shopify, for their part, protected all 1-to-1 catch-ups regardless of the recurrence during their meeting purge. That is because one of the primary factors of meeting quality is meeting size. For today, we will focus in on big meetings, why they tend to be bad meetings, and tactics for building better meetings - both big and small.

‍

The Big Meeting

In Shopify’s meeting purge, 1-to-1 meetings were left untouched. This is critical from a management and development perspective. Yet, even for peer-to-peer brainstorming the holdout makes sense. Most people I talk with love their 1-to-1 meetings, no matter the context. They are incredibly dynamic and allow for lively debate to rapidly align on context between the two sides. New ideas can be presented and discussed so that each person can fully understand the other. They are also productive and usually cover multiple topics. Because context can be quickly and fully clarified, topics do not take long to cover. This allows team members to tackle a number of items, one after another. On the other end of the range, large team meetings feel slow. Even a simple task like introductions can easily wipe 10 minutes of time off of the session. Debate, if it happens at all, is often incomplete. We all know what is happening here - more people in the same time constraint means less time can be allotted to each conversation.

However, we likely do not realize exactly why this happens or just how much potential is being lost. To realize how quickly these conversations degrade we can look to combinatorics. Combinatorics is a broad area of mathematics covering - at the very least - the counting of finite structures or arrangements, as well as the construction and optimization of these arrangements¹⁰. Combinatorics shows up in physics, evolutionary biology, and now in meeting attendance.

One phenomena we can describe from this field is combinatorial explosion, the rapid growth in a problem’s complexity as structures are added¹¹. In our case, we could imagine the simplest arrangement as a 1-to-1 to cover a specific topic. I modeled this simple arrangement below. In this set-up, all information sharing is direct. This arrangement is highly efficient, understandably effective, and deeply beloved.

The above 1-to-1 might uncover a particularly important topic with the takeaway of reviewing the go-forward plan with a subject matter expert to get a more nuanced and technical point of view. When this person joins, you each need to make sure that the expert understands your individual points-of-view. You each also need to ensure that you are drawing the same conclusions from the expert’s feedback. We can represent this meeting with the arrangement below. Note how we only added 1 person but added 2 connecting threads.

‍

If the initial 1-to-1 had required an even more robust discovery meeting follow-up, the number of connecting threads could quickly add up. With the same approach we see that a 4-point meeting has 6 connecting threads. A 6-point meeting has 15 connecting threads. An 8-point meeting has 28 connecting threads. We can model this for any meeting size. For Shopify’s case, they restricted meetings of more than 50-people to a 6-hour block on Thursdays. That 50-point meeting would have 1,225 connecting threads.

‍

‍

This is where the real problem begins. A “connecting thread” has no meaning. We must try and apply a legitimate “real-world” value to that variable placeholder. For example, we might assign the value of “fully debating and aligning on an idea” to a connecting thread and forecast that to require 5 minutes in a 1-to-1 meeting if things go smoothly. For a 4-point meeting with 6 connecting threads, that equivalence is 30 minutes of debating and aligning. For an 8-point meeting, that would be about 2 hours and 20 minutes. In the Shopify scenario of 50 people, we would need about 4 days and 6 hours to fully hit every connecting thread. If you have heard of Jeff Bezos’s “two-pizza rule” - the idea that no team or meeting should need more than two pizzas to feed everyone - it is probably starting to sound pretty good¹².

‍

The Bad Meeting

Complaints about meetings are endless. But, I have never heard anyone’s primary complaint be that a meeting ran for more than 4 days. Rather, senior leaders highlight low engagement from their team during meetings. Mid-level managers vent over the lack of debate and feedback that they were hoping to get from their topic sections. Junior team members call out that a majority of the content is not relevant to them. When we consider the combinatorial results of increasing a meeting size, the behavioral outcomes we see and their impact on meeting value start to make a lot of sense.

‍

Self-Regulation

You never hear complaints about 4-day meetings because they do not exist. Every meeting is capped at some time constraint. For big meetings, this is generally an hour. There is an inherent understanding that every aspect of every topic will NOT get full coverage. Because of this, everyone is playing a constant game of evaluating how they should contribute in order to maximize the value of the meeting time. Some leaders assume that their teams are handling that evaluation exclusively on the merit of the ideas they have to share. Those leaders are wrong.

Humans are social creatures. We have millions of years of evolutionary programming that drive our ability to read and respond to social cues with limited effort. In fact, multiple studies showcase our ability to solve optimization problems and puzzles once they are wrapped in a social context. The Wason selection task is among the most famous examples¹³.

In the original study, fewer than 25% of Wason’s participants identified the correct solution. That result has been replicated in a number of studies. However, the same problem can be reconstructed. When it is reformatted, more than 75% of participants get the answer correct. We have to frame it as a social problem¹⁴.

There are a few factors at play that influence task performance - including rule familiarity and general intelligence - but rules that represent social interactions are consistently solved at higher rates. It is reasonable to expect that this social analytical capability translates beyond the world of research study and experimentation.

For our junior team members, it only takes a few flips of the card to understand how much they should be looking to contribute in a big team meeting. We can create another instance of the Wason task to show how this conclusion could be realized.

Just like in the previous tasks, with only 2 flips we can learn the rule for whether we should engage in the meeting. If we are a junior team member and time is short, we should not speak up and instead just have a conversation after the meeting. We could imagine a Wason task where the colored cards represent engagement received for a comment. A senior leader will always get some type of response; junior team members, sometimes. Humans are incredibly sensitive to social cues. It only takes a few scenarios and a few card flips to learn a rule set that encourages us to self-regulate our engagement to low or nonexistent levels. Add a little bit of imposter syndrome to the mix and it is surprising that most team members engage at all.

‍

Scope Limitation

Another common criticism of meetings comes from managers who want to foster debate in meetings. Ideally, a big meeting should gather a variety of functional experts who can contribute alternative opinions and, through their pushback on an initial idea, make it better. In reality, we get only a few voices sharing slight variations on the initial thought. The cause becomes clear if we think back again to the initial mathematics.

Let’s use the 8-point meeting as an example and assign some functional titles to the different nodes. If we had a fully cross-functional debate, we might see the arrangement light-up with activity, like so. The conversation may be very valuable but we can also expect for it to take a while.

More realistically, the conversation will not stretch to touch every downstream impact. For example, a new marketing campaign could touch the human resources function if we need to staff up or bring on free-lance support. For the sake of time and simplicity, the debate may simply exclude that consideration. By removing the HR components of the idea, we reduce the number of connecting threads down from 28 to only 21. This intentional scope limitation seems practical and desirable.

But remember, everyone plays a role in maximizing the meeting’s value in the given time constraint. And we have already discussed how people might opt themselves out. Here, we could see the Software Development team decide that their counterpoints are too narrow in scope and stay quiet. The Client Services team might have concerns on staffing - which would require HR involvement - but opt out because they know they will be pushed to help deliver the campaign regardless. The Business Operations team may think they have a full understanding of the problem and choose to not take time to confirm that…only to later find that they disagree with some key tenets of the new campaign. Suddenly, our arrangement looks incomplete.

This has turned into a 4-point meeting, which fits into 30 minutes. Without thoughtful evaluation, the topic may be seen as a success because the campaign was discussed within the allotted time with all stakeholders present and no disagreements were raised. We must realize that there could be danger hiding in that perceived alignment.

‍

Shared Relevance

The third common criticism comes from more junior team members, who often call out that topics from meetings are usually not relevant to them. In most cases, a big meeting is typically going to cover cross-functional topics like weekly business performance or a tactical strategy review. While these will engage different team members at different points throughout the meeting, this meeting is usually designed by and for one person - the general manager. This can be an issue because leaders rarely ask their team for feedback on a meeting’s value. Instead, they evaluate whether the meeting was relevant and productive to them and assume the same holds true for everyone. This is a blind spot.

Because their scope of responsibility is to manage across activities, all topics will be relevant to the GM, which creates a false sense of relevance to the rest of their team. For the other functions and team members, their work has more overlap with some functions than with others.

We can visualize that variable relevance like below, with a bolder line representing more relevance than a thinner line.

For some stakeholders, like the General Manager and the Sales team, a majority of the content has high relevance. This creates more opportunities to engage with any topics that come up and to be a contributor. After a meeting, these teams with a high amount of shared relevance will likely evaluate the meeting as more productive.

For other teams, like Software Development or Human Resources, there are only 1 or 2 paths to them with a high degree of shared relevance. We can think of these line weights as probabilities. If a topic being discussed is focused on Client Services, there is a low probability that Software Development will have shared relevance and have anything to contribute.

Those team members will naturally do some expected value calculations to decide whether it is more productive for them to pay attention or multi-task. In most cases, the sure-fire nature of even low-value tasks will drive greater returns than the low-probability of maybe sharing some highly relevant context during a debate. Further, this can create negative feedback loops as the team’s multi-tasking will cause them to lose track of the meeting conversation and miss reasonable opportunities to contribute. This further lowers their perceived probability of adding value, leading them to check out even more.

As with any probability calculation, we can combine the probabilities into a single value for any functional team by multiplying the “values” of each connecting thread together. For the GM who has a 95% chance of finding relevant information from the topics of any function, the overall chance that the meeting will be useful to them is =0.95*0.95*0.95*0.95*0.95*0.95*0.95 or right about 70%. Those are pretty good odds for the GM that they will find it useful. For our Software Development team that may look like =0.3*0.95*0.5*0.3*0.95*0.3*0.3 or about 0.4%.

Note that even for the GM, increasing the meeting size decreases the probability that they will find something of useful relevance. We could model this for a 4-point meeting and get a probability of relevant information of =0.95*0.95*0.95 or ~85%.

‍

The Better (?) Meeting

As meetings become bigger they become less engaging, more limited in the scope of conversation, and feel less relevant to more members of the team. The best advice I can give you is to be extremely diligent on the size of your meeting attendance list and, wherever possible, have fewer and smaller meetings.

Realistically, this is not always feasible. In those cases, we can try some off-the-wall tactics to at least control some of the adaptive behaviors that our teams adopt in response to big meetings. Tactically, we want to highlight the concept of ‘scope limitation’ in order to try and improve engagement by highlighting relevance and sharing room for team members to disengage. This admittedly looks a lot like pre-sharing an agenda. But like I said at the outset, those ‘good meeting’ tactics are not enough to create a good meeting. Here are some key differences to add on top.

‍

1. Define Rotations - For each agenda item, also include who is expected to participate in the conversation. Only include teams where the probability of shared relevance is high or very high. Set the expectation that the teams in the On rotation should be fully engaged. But, also set the expectation that teams in the Off rotation are free to fully disengage.

The goal is to improve engagement by creating time to be disengaged from the conversation. As we know the standard behavior in big meetings is to multi-task. However, the research on multitasking shows that we are highly ineffective when doing this and bouncing between cognitive processes. With rotations you can remove the need to multitask between focusing on the conversation and pushing other pieces of work forward. The Off rotations are encouraged to put their full focus somewhere else. Inversely, the On rotations are able to fully engage in the conversation. They already know that the probability of being able to contribute is high and that the topic is likely worth their focus. Additionally, they do not need to feel distracted because they know that they will be free to disengage at the appropriate time.

‍

2. Include Time Forecasts - A standard meeting agenda will list the order of topics. It is generally assumed that you will cover the next item as the previous one wraps up. Instead, forecast how long each topic should take using time-and-a-half forecasting to allow sufficient time for engaged conversation. If you are running over on a topic, it is up to the leaders to make a call on whether they should move to the next topic or keep running. When making that decision, KEEP IN MIND whether your primary meeting KPI over the long-term is engagement or timeliness and consider what message you may be sending based on that decision. If you choose to run long, make sure to update the Off rotation of the updated time forecast.

If the topic runs under the forecasted time, take a break. Let people relax, grab a water, talk about the weekend, or have more sidebar conversations about the current topic. Do not move on to the next section until the previously defined time. Your aim in doing this is to reaffirm the rotations tactic. If you ignore the defined time blocks, you are implicitly telling the teams that they need to pay attention at all times anyways. This sends everyone back into a low-impact multitasking mode.

‍

3. Highlight All-Team Conversations - If we are covering tactics for improving the productivity of big meetings, the prior two ideas might, sneakily, sound like just scheduling smaller meetings. However, one thing a smaller meeting can never replicate is that a big meeting has the full team present and accounted for. Sometimes that is exactly what you need.

The General Manager or team leader will likely be in every On rotation. They are participating in every conversation and watching new ideas or problems develop. In the meeting they have the authority to, with sufficient fanfare, update the rotations and add teams to or drop teams from the On rotation. They may even decide to call the full-team in to the conversation.

To do this effectively, the leader has to do two things. First, they need to make the change marker abundantly clear. They have to grab the attention of those in the Off rotation who, by design, should be pretty intently focused elsewhere. If the changeover is too subtle, the team will again start to infer that they need to pay attention even when on an Off rotation. Second, they need to make this uncommon. Remember that we are balancing short and long term gains here. Switching the On rotation assignments on the fly can be impactful to the short team, but over the longer term will diminish the value of the Off rotation and risk multitasking and lower engagement.

‍

The Value of Learning

I will share one final thought to close out. One final argument in favor of big meetings is their role in development. The concept contends that seeing the interaction between more senior team members in a big, cross-functional meeting can help educate junior team members on how the company thinks through strategy. Additionally, the presentation reps that come from big meetings can help promote the development of presentation and persuasion skills. I fully agree.

However, make sure that this value is clear to everyone. If it is important that junior team members learn from big meetings, tell them that you expect them to pay attention and take note of questions that they have and lessons they learn from the time. Also, tell managers that you expect them to schedule 1-to-1 time with their reports immediately after meetings to answer those questions while the topics are freshly top-of-mind. If this cannot be done, I would challenge how important this meeting is as a development tactic.

Also, make sure that this is not the only tactic that your business uses for development. While team meetings can be an effective medium for practicing presentation skills, there are certainly other ways to create those opportunities that are far less expensive. Similarly, structured L&D sessions that describe the company’s strategic principles and give tactics for working through business problems are great to do as part of ongoing training. This investment in dedicated training time will help ensure that you get the most value out of the on-the-job education that a meeting can provide.

‍

References

Combinatorial Explosion & The Big, Bad Meeting