How long was that meeting?

What’s the longest meeting you’ve ever sat through?

Mine was only fifteen minutes long.

Wait, what?! How does that make sense?

I’ll tell you the story, but first, we’ll talk theory…

Let’s suppose you have two meetings. The first is one on one, but it lasts for three hours! The other lasts just thirty minutes, but there are twelve participants. Which is longer?

Trick question. Hint: they both take exactly the same amount of time – SIX hours!

Six, because the first meeting took three hours of time from two participants' schedules (3h x 2 = 6h) and in the second, thirty minutes from twelve participants (30m x 12 = 360m = 6h).

That’s basic math, of course, but if you’re like me, it’s still not that intuitive. We all know that meetings are expensive in terms of time, but it’s easy to forget that the more participants join, it doesn’t become marginally worse, but dramatically so. You need to remind yourself of it, because it has big implications.

To accomplish a task, it may be much more efficient, for example, for you to spend a whole day in meetings, rather than take fifteen minutes of your whole team’s time with a short discussion (or e-mail thread), depending on the size of your team.

But especially if that’s not immediately obvious to you, it may be very tempting for you to go with the least efficient option by asking everyone for fifteen minutes, because for you it saves so much time, and after all, who can’t spare just fifteen minutes? It’s a classic tragedy of the commons, except that in this case, the common good is time – not money, services, or land.

All else equal, even with the right time math, the bias should still be against larger meetings, because when one person speaks, everyone feels an urge to respond. It turns out that there’s hidden math here, as well. Time for another example…

Suppose there are two meetings: one with two participants, the other with eight. Let’s assume that both conversations (including all participants) are equally provocative, such that if both begin with one participant speaking for the first five-minutes, the other participant(s) want to reply for five-minutes. Which meeting lasts longer? The larger one: (1 x 5m) + (7 x 5m) = 40 minutes long instead of (1 x 5m) + (1 x 5m) = 10 minutes long.

What if our assumption is wrong? Perhaps smaller conversations are just more stimulating. Or maybe in larger ones, even if everyone wants to give their full five-minute reply, they will restrain themselves out of consideration of their colleague’s time, pressure from social norms, etc. and cut it short, by half.

So, let’s dispense with our old assumption and replace it with a new one: smaller conversations are twice as provocative and larger conversations are half as provocative as we said before. What do we get now?

SmallerMeeting = (1 x 5m) + (1 x 10m) = 15m
LargerMeeting = (1 x 5m) + (7 x 2.5m) = 22.5m

There’s your proof. Even though the larger meeting, in this example, has four times more participants, we compensated for that by halving its response time and doubling that of the smaller meeting. Convinced?

In case you’re not, let’s do a little more math to prove the point. After all, maybe this example is rigged. We did, after all, give the larger meeting four times more participants. So shouldn’t we be dividing its response-time by four? You end up with this:

SmallerMeeting = (1 x 5m) + (1 x 5m) = 10m
LargerMeeting = (1 x 5m) + (7 x 1.25m) = 13.75m

Or, if you do the opposite, and multiply the response-time of the smaller meeting by four, you end up with:

SmallerMeeting = (1 x 5m) + (1 x 20m) = 25m
LargerMeeting = (1 x 5m) + (7 x 5m) = 40m

I’m no math whiz, so maybe I’m wrong, but this exercise seems to indicate that you should have a bias towards smaller meetings because when you factor in the human reality that people want to respond to each other, they are more efficient.

It gets worse, as if it weren’t bad enough. If you just take a moment to think about it, what I’ve been calling ‘provocativeness’ - our inclination to respond to each other - in most situations is best modeled by either an exponentially decaying or multiplying function. Let’s put this in human terms. A normal conversation, between two people, goes like this: I talk, you respond, then I respond to you, you respond back to me, and so forth. Conversations can either wax or wane. When they wax, the response time in each back-and-forth cycle is either stable or increasing. When they wane, it decreases.

What this helps us recognize is that the SmallerMeeting and the LargerMeeting in the example above rarely end after the first cycle of conversation. Each set of replies trigger a new set of replies of waxing or waning length. Here’s my best attempt to model that dynamic:

WAXING by 1.5x for five cycles…

SmallerMeetingCycleOne = (1 x 5m) + (1 x 5m) = 10m
SmallerMeetingResponseTwo = 10m + 1.5 x (1 x 5m) = 17.5m
SmallerMeetingResponseThree = 17.5m + 1.5 x (1 x 7.5m) = 28.75m
SmallerMeetingResponseFour = 28.75m + 1.5 x (1 x 11.25m) = 45.625m
SmallerMeetingResponseFive = 45.625m + 1.5 x (1 x 16.875m) = 70.94m
1 hour and 10 minutes

LargerMeetingCycleOne = (1 x 5m) + (7 x 5m) = 40m
LargerMeetingResponseTwo = 40m + 1.5 x (8 x 5m) = 100m
LargerMeetingResponseThree = 100m + 1.5 x (8 x 7.5m) = 190m
LargerMeetingResponseFour = 190m + 1.5 x (8 x 11.25m) = 325m
LargerMeetingResponseFive = 325m + 1.5 x (8 x 16.875m) = 527.5m
8 hours and 45 minutes

WANING by .5x for five cycles…

SmallerMeetingCycleOne = (1 x 5m) + (1 x 5m) = 10m
SmallerMeetingResponseTwo = 10m + .5 x (1 x 5m) = 12.5m
SmallerMeetingResponseThree = 12.5m + .5 x (1 x 2.5m) = 13.75m
SmallerMeetingResponseFour = 13.75m + .5 x (1 x 1.25m) = 14.375m
SmallerMeetingResponseFive = 14.375m + .5 x (1 x .625m) = 14.6875m
15 minutes

LargerMeetingCycleOne = (1 x 5m) + (7 x 5m) = 40m
LargerMeetingResponseTwo = 40m + .5 x (8 x 5m) = 60m
LargerMeetingResponseThree = 60m + .5 x (8 x 2.5m) = 70m
LargerMeetingResponseFour = 70m + .5 x (8 x 1.25m) = 75m
LargerMeetingResponseFive = 75m + .5 x (8 x .625m) = 77.5m
1 hour and 15 minutes

Guess what tickles me about this? When I crunched the numbers, I didn’t know what to expect, but I feel like this is fairly indicative of my actual experience. When I’m in a long meeting – like an advisory board session, for instance – with eight others and we’re talking about something that is very ‘provocative’ (interesting, controversial, substantive, important, etc.) it feels like the conversation expands rapidly, then, just when, from a conversational stand-point, we’re barely getting starting, at around the third round of responses, we’ve spent over three hours and we’re out of time! It’s almost as if, to get it all out, we would need an Entmoot!

“Since the Ents' language is so descriptive and ‘unhasty’, an Entmoot can take a very long time. In the Lord of the Rings, the Ents met for three days to decide if they should go to war, which is considred quick for an Entmoot.”

Conversely, if I find myself in a room dealing with a relatively less ‘provocative’ subject (not as complicated, not as controversial, not as critical), even if it starts winding down from the beginning, it can take over an hour to just wind-down. I think that last model – LargerMeeting when waning – is most similar to the sorts of meetings I find myself most often: scheduled for an hour, running over-time by fifteen minutes, when most of what really needed to be discussed could have been handled in fifteen minutes (in a SmallerMeeting when waning model).

It’s not that simple of course and other factors come into play. Perhaps by letting everyone participate in a big meeting, there’s some added benefit from distributed awareness, consensus, or lots of different perspectives. In a medium-sized meeting, there may be benefits to having some of the key-stakeholders present for making a decision. On the other hand, by keeping it small, there’s professional intimacy, you’ll both feel real ownership of the output, and you can concentrate more easily on co-creation.

Also, spliced schedules suck. Certainly I am not the first to observe that when your meetings are spread out, either spatially or temporally, it gets harder to get stuff done in-between. It’s overhead: the higher it is, the greater the tax on efficiency.

Last but not least, some people’s time is more expensive than others, so you need to take that into account as well. This isn’t just about pay-scale, either. It’s possible that the most junior person on your team happens to be working on what is, for the moment, the single most critical issue right now, and if he becomes a bottle-neck because he got dragged into a meeting, the delay could be very expensive. For example, if all the big-ticket engineers have finished their coding for the day and they go into one last meeting, maybe the young QA engineer should sit it out, if a build is about to be pushed and bugs need to be found and resolved.

Even so, calculate the real length!

To wrap up, three simple tips:

First, for every meeting next week, do the math on the real time spent, and share it with your team. Awareness is the first step.

Second, encourage anyone, at any time during a meeting, to interrupt, excuse themselves if they feel they are no longer needed, or ask whether the discussion currently in-motion is best-served by everyone being present, or whether it should be broken up into one or more smaller meetings.

Third, encourage other healthy meeting habits. Maybe, before someone calls a meeting, a written agenda should be circulated. That way, when a response, to a response, to a response has side-tracked the conversation way off course, there is an agreed-upon anchor to return to.

Back to the story behind my longest meeting…

When I interned at Google, I used to sit in on internal conference calls. Most were fairly brief. Some had dozens of participants, others had literally hundreds, and a few were “all-hands” meetings for whole departments or even, occasionally, the entire company. Those big ones, even if they had lasted for just fifteen minutes, were certainly longer than any meeting with an individual or small group I’ve ever been in, or ever will be.

It’s just math.

Do the math!

My colleague Joe Nangle (@joenangle on Twitter) just asked that we institute the mandatory-agenda policy in Tip #3. Smart guy!

Today, I almost didn’t write a blog post, even though I’ve kept up a nice streak for almost a week. I assumed nobody would notice or care, but I was wrong. I dedicate this post to my friend David Matthews (@davidolski on Twitter) from Sponsorfied, who reminded me tonight of the value of consistency, even if it’s just publishing one paragraph, and gave me the encouragement I needed to keep writing a post a day. Thank you, David: I started this post fully intending it to be one paragraph, I promise!