Mathematics of Meetings

My profession is to go to meetings and write documents, I often tell my 6 year old. A lot of work happens at meetings. So if one wants to accelerate progress at work, getting a new product launched, ramping up sales, finishing a report, whatever, there are two ways to go faster: make each meeting more effective, and/or have the next meeting soon.  (I got this idea while swimming, where there are only two things that affect your lap time: stroke length (how far you propel yourself with each stroke) and stroke rate (time between strokes).)

Many articles and techniques have been written about the subtle interpersonal techniques around how to have an effective meeting (stroke length).  Not so much has been written about meeting frequency (stroke rate).

Fortunately, meeting frequency is simple math: reduce the wait time to next meeting, and your initiative or organization moves faster. The target metric (or Key Performance Indicator, KPI):  Average Time to Next Meeting.

So one Friday night, I pulled out my trusty loved-hated Excel and tried to model out what are the key variables that affect this time-to-next-meeting. I was surprised to find a few very important variables that massively slow an organization.  Here they are:

1. The number of people needed at the meeting.

We all know that more people means slower progress, but I was shocked to see in numbers how much a few extra people delay the next available slot — like Covid, it’s exponential.  Say I have  hour-long meetings randomly throughout an 8 hour work day, and usually spend half my day (4 of the 8 slots) in meetings. I want to find one slot where both of us are free.  Maybe at 9am I am free, but not my colleague; 10am my colleague is free but not me; 11am we are both in meetings; and noon we are both free.  In the model, it takes us an average of nearly 4 hours to meet up — so still same day, no biggie.  Add a third person, and on average it will take us 10 slots to meet up, so by tomorrow morning.  Add a fourth, and now it’s 20 slots (within 2-3 days); to have 9 people meet, it’ll take 534 slots, or nearly 10 weeks.

Top tip to speed up work: have max 3 people in a team that need to meet.

2. What proportion of each day a person spends in meetings

I assumed here all of us had already booked 50% their day with meetings (4 of 8 slots).  Some people rarely have meetings, and can be scheduled in any day.  But if you are trying to schedule in someone whose main job is to have meetings, for instance a manager or senior executive, they may spend 80-90 or even 100% of their time in meetings.  You don’t need an Excel model to figure out that scheduling a meeting with someone whose calendar is already 100% full for weeks means you won’t be able to schedule a meeting at all.

Top tip: cut down the number of meetings you commit to, and avoid including people who are always in meetings (such as senior managers).

2b. …especially when >85% of your day is in meetings

You will know either from trying to fly during the (pre-pandemic) Christmas holiday, that when people and planes are operating near 100% of capacity (or really anywhere above 85%), your waiting time explodes from minutes to hours.  When we want to go faster in business we tend to load people up to 100% (or 150%) of the time they have;  loading people’s calendars more than 85% with meetings hits a real tipping point (Here’s a picture from queuing theory) — you end up needing to wait a LOT longer to have that next meeting.

The fix: resist adding more work until you and the many stakeholders involved have finished the first. Celebrate those who operate at less than 85% of their capacity — who have some down time.  Culturally, it can be hard to tell a boss (or boss’s boss) you’re too busy for them; but the fastest teams are the best at pushing back on taking on more work.

3. How many slots there are per day.

There are 2 ways of increasing the number of slots per day : longer working hours, and/or have shorter slots.  If you are looking for a slot, it’s a lot easier to half the meeting length (e.g. from 1 hour to 30 minutes) than to double the number of hours worked (from 8 hours to 16 hours).
Tech teams famously use the 10 minute standing meeting to address this problem.  In general, I’ve often wondered Outlook/Google calendar had 20, 40, 60 minute default meetings rather than 30 and 60 minute ones, if we’d all meet up faster.  Working remotely has helped put more slots into a day — no more travel time to and from meetings. But days of back-to-back virtual meetings mean we, as humans, need longer slots: if you have too short, task-focused meetings only, you destroy human connection; and often don’t let the important stuff that really matters come to light.

4. How correlated their schedules are.

In my model, I’ve randomly assigned a calendar slot to empty or full — there’s no correlation between my calendar and my colleague’s.  But that’s not usually the case.

One way teams solve the Time to Next Meeting problem is by coordinating schedules well in advance and on a recurring basis — the standing Tuesday 9am team meeting, the monthly senior management meeting, the annual offsite conference. Of course, if your stakeholders’ calendars are now full of recurring standing meetings, go back to top tip 2 above!

I’ve also assumed that calendars can’t change. But often they do: you can propose a time and the other person can move something around to make it work.  How helpful would Outlook, Google Calendars, Calendly and other calendar programs be if they knew where there was some flexibility in the slot, and could adjust others’ calendars for you!

One common ‘calendar correlation problem’ large corporates face is using specialists : legal, financial, technical, central services, etc. Often their calendars are truly uncorrelated to the teams they support. The main fixes I have found are switching to one-to-ones with the experts, giving up on meetings entirely and switching to email, using specialists in a very short timeframe (e.g. all within the same week when they might have a lull from other projects) or abandoning specialist support entirely.  If you have other solutions, let me know.

5. How much the team’s working hours overlap.

Having worked in global jobs spanning multiple time zones my entire career, I know working across time zones kills any hope of that next meeting happening quickly.  I chose to live in London in part because it straddles the Americas-Europe-Asia time span.  But if you need a team with people from London, the US (hopefully not California), Bangalore and Singapore, you likely have 1 (awkward) hour a day to meet, at best — 7 am California (preferably not a Monday) / 3pm London / 10 pm Singapore (preferably not a Friday).

It’s not just time zones — it’s holidays too.  My first summer working in Europe, I was advised to ‘make sure you get your deliverables done by June, before the July and August holiday period.’ Having moved to London from China, where all the nations holidays were synchronized around (then) 3 major holiday weeks, and in America, where people didn’t take more than a few days holiday at a time, it was a learning curve: some schools (and their parents) break for a few weeks holiday in July, others in August, Germany has a week long holiday sometime in October, others take a few weeks off for Eid, in addition to the Christmas, New Year and Chinese New Year holidays I already had learned to work around.  In my model, I didn’t assume any holidays — once you add those in, the time-to-meeting explodes.  The solutions I have seen are to synchronize shut downs (e.g. everyone takes off the week after Christmas), reduce participants from teams that have different holiday schedules, pre-plan projects with the holiday calendar in mind and ask people to work during their holidays (ouch).

So that’s it: it’s just math.

It’s quite amazing that there’s a single metric that you can measure to figure out if you work in a fast organization/team/project, or a slow one: ask your colleagues how long it takes on average to book the next meeting. Is it same day, next day, 2-3 days, >1 week, or >2 weeks?

Not everyone wants a fast organization, and a lot of mistakes are made by going too fast. There are good reasons for having a slower organization, where the tradeoffs are worthwhile: maybe you make huge, multi-year investments, with safety critical operations, and massive risks.  In that case, it’s worth it.

But if you are in a business or organization which doesn’t have these constraints, especially where the market is changing quickly or you have aggressive growth targets to hit, then consider a few ‘quick’ fixes:

  • Delegate power and authority to small (3 person) teams
  • Keep everyone working in the same time zone, on the same correlated schedules; avoid jobs or reporting lines that span time zones
  • Keep everyone at least 15% below what they can do (think: a day free a week)
  • Shorten the default meeting time to 20, 40 or 60 minutes
  • Use specialists with uncorrelated schedules sparingly, in one-to-ones, in bursts, or by email
  • Don’t require all decisions or input to be made, live, by people whose job it is to primarily attend meetings

Does this resonate with you? Any other variables you noticed? Any other fixes you’ve found that work?

And if you’ve made it this far, let me know — it means I found someone else who finds it interesting to think about the mathematics of meetings!  We should be friends.



If you enjoyed my writing, you may enjoy my recently published novel Turning Forward, available in eBook and paperback on Amazon