urmeetingcost.lol, What Actually Drives the Cost of a Meeting
Most organisational advice about expensive meetings reaches for the same lever: make them shorter. Twenty-five minute slots, no-meeting Wednesdays, the dreaded "this could have been an email." It is well-meaning and, on the data, mostly the wrong lever. The dominant driver of meeting cost is not duration. It is who is in the room.
The problem
Meetings have a real financial cost that almost no one calculates and almost everyone underestimates. Six senior people in a room for an hour is not "an hour." It is a number that, if you wrote it down, you would probably not have called the meeting in the first place.
I built urmeetingcost.lol to make that number visible. Type in your attendees, their salary band, the meeting's duration, and the tool returns the salary-loaded cost. The interesting part came later: users were entering enough real-shaped data that a follow-on analysis could ask a more useful question, namely which input variable actually moves the cost the most? Most people assumed it was duration. The data said otherwise.
The approach
A small, deliberately scoped public tool plus a statistical follow-on against the inputs people had given it.
The tool. React and TypeScript front-end on a Flask backend with SQLite. Attendees can be entered by job title (auto-resolved to an indicative salary band) or by raw salary. Duration is in minutes. The tool returns a per-meeting cost, an annualised cost if the meeting recurs, and a small comparison chart against a benchmark of typical meeting shapes. No login. No tracking. Just the number.
The data. Anonymised inputs from the tool gave a working dataset of meeting shapes: attendee counts, salary mixes, durations, and resulting costs. Skewed toward the kinds of meetings people are most worried about, which is exactly the population the analysis wanted to characterise.
The analysis. Three methods, used to triangulate rather than to mistake any one of them for the whole answer.
- Correlation studies. Pearson and Spearman correlations between each input variable and total meeting cost. Salary correlated strongly. Attendee count correlated moderately. Duration correlated weakly. The ordering surprises people.
- Random forest models. Feature-importance analysis confirmed the correlation finding: salary dominated, attendee count second, duration third. The model gives the answer regardless of input distribution assumptions the correlation tests rely on.
- Interaction terms. Salary and attendee count interact strongly. A roomful of executives is expensive in a way that is more than additive. The cost amplifies because each attendee in a high-salary band amplifies the others.
What the analysis found
The headline result: salary is the silent heavyweight. A thirty-minute meeting with two executives can cost more than a ninety-minute meeting with five junior analysts. The "shorten the meeting" reflex saves real time but does not move the cost dial much, because the cost dial is mostly attached to who is sitting in the room.
The compounding finding is more useful for the meeting-organisers who care: there is a sharp interaction between salary band and attendee count. Once you cross a threshold of senior attendees, every additional senior attendee makes the meeting disproportionately more expensive, not just incrementally so. The practical advice that falls out is not "make meetings shorter." It is "audit the invite list before sending it, especially when the invitees are senior."
Evidence
- Salary band identified as the dominant cost driver across correlation, random forest, and interaction-term analyses
- Attendee count identified as the secondary driver
- Duration shown to have the weakest correlation among the three inputs
- Salary × attendee-count interaction term significant and large, indicating amplification rather than addition
- Practical recommendation supported by data: invite-list discipline outperforms duration discipline for cost reduction
- Public tool live and free to use at urmeetingcost.lol, with no tracking and no login
The work is small. The lesson sits in the gap between the obvious lever (shorten the meeting) and the actual lever (invite fewer expensive people). Decision support, in the smallest possible form.
← kipjordan.com