What meetings actually do to productivity — the honest research.

Most arguments about meetings and productivity rest on surveys — surveys of executives, surveys of senior managers, surveys of knowledge workers asked to estimate their own focus loss. Useful, but they describe perception, not effect. The peer-reviewed work that actually measures something has been quietly accumulating, and it’s a smaller, more careful body than the LinkedIn discourse suggests.

This piece is a synthesis of that smaller body. What’s been measured. Where the honest gaps live. What a leadership team is justified in concluding from the evidence as it stands today — and, just as importantly, what it isn’t.

Interruptions are costly, and meetings are scheduled interruptions

The most robust productivity-relevant finding adjacent to meetings isn’t about meetings themselves. It’s about interruptions. Gloria Mark, Victor Gonzalez and Justin Harris’s 2005 CHI paper, “No Task Left Behind?”, measured average recovery time after a workplace interruption at roughly 23 minutes — the now-canonical figure — and observed that fragmented workers tend to compensate by working faster, not by working less.

A scheduled meeting is a scheduled interruption. The productivity cost of a 30-minute meeting is not 30 minutes; it’s 30 minutes plus the recovery on either end, plus whatever was queued to begin in the slot the meeting now occupies. That tax is real, measured, and mostly invisible on a calendar.

Video meetings cost more cognitively than in-person ones

Stanford’s Jeremy Bailenson published the most-cited theoretical account of why video calls feel disproportionately heavy in the APA’s Technology, Mind, and Behavior in 2021. Peer-reviewed, primary source, four distinct mechanisms identified: prolonged close-up eye contact, the cognitive load of watching yourself on a screen, reduced physical mobility while on camera, and the load of decoding non-verbal cues through a lower-bandwidth channel.

The paper is a theoretical argument, not an effect-size measurement, and treating it as either is a mistake. What it does give you is a defensible explanation for why a 23-hour-a-week video calendar — roughly the top-quartile figure Microsoft reported in 2023 — costs more in energy than the equivalent in-person calendar would have ten years ago. Same hours; different ledger.

The COVID natural experiment said: work expanded

The cleanest large-N measurement we have on what changes when meeting load shifts is Evan DeFilippis, Stephen Michael Impink, Madison Singell, Jeffrey Polzer and Raffaella Sadun’s 2022 paper in Humanities and Social Sciences Communications. The dataset: 3,143,270 users across sixteen metropolitan areas in North America, Europe and the Middle East. Method: event study around lockdowns.

What they measured. Meetings per person per day rose 12.9%. Attendees per meeting rose 13.5%. Average meeting length fell 20.1%. Total time in meetings per day fell 11.5%. The length of the average workday rose 8.2% — about 48.5 minutes — and the increase was durable, not transient.

Most pieces that cited this dataset led with “meetings got shorter.” The reading they missed: time spent in meetings actually went down; the workday went up; and the longer day stayed. Whatever displaced the meeting time wasn’t reclaimed for life. It re-emerged as more email, more async messages, more late-evening work — and more total hours.

This is the closest thing the literature has to a measured productivity effect of changing meeting load at scale. The answer isn’t encouraging. Reducing the calendar without also reducing the underlying workload tends to re-route the cost rather than remove it.

More sync didn’t make collaboration tighter

The Microsoft Research / UC Berkeley team published an even larger, more carefully-controlled paper in Nature Human Behaviour in 2022 — Longqi Yang, David Holtz, Sonia Jaffe and colleagues, n=61,182 Microsoft employees observed across the first six months of 2020. The question they asked was specific: what did the shift to firm-wide remote work do to the structure of collaboration?

The finding ran against intuition. Collaboration networks became morestatic and siloed, not less. Communication shifted toward asynchronous channels. People talked to the people they were already talking to, and stopped talking to the people they used to bump into. Adding sync time was not the connective lever it tends to feel like — the connective tissue of an organisation isn’t made of scheduled meetings, it’s made of the ambient lower-formality contact that scheduled meetings tend to crowd out.

Meeting-free days: directionally encouraging, methodologically soft

The most-cited recent piece on meeting-load reduction is Benjamin Laker, Vijay Pereira, Pawan Budhwar and Ashish Malik’s January 2022 article in MIT Sloan Management Review. They surveyed 76 large companies — each with more than 1,000 employees and operations in 50+ countries — that had introduced one to five no-meeting days per week.

The headline finding: introducing meeting-free days correlated with self-reported gains in autonomy, engagement, communication and satisfaction. The strongest results landed at three no-meeting days per week. About 47% of the studied firms ran two; 35% ran three; 11% went to four.

The caveat the paper itself names: this is self-reported survey data. Perceived productivity is not measured productivity. The case for meeting-free days as a directionally good intervention is real; the case for any specific percentage gain is not. If you’re bringing this evidence to a leadership team, bring both halves.

What the meeting-science review says we should actually do

Joseph Mroz, Joseph Allen, Dana Verhoeven and Marissa Shuffler’s 2018 review in Current Directions in Psychological Scienceis the single best entry point into the peer-reviewed meeting-science literature. They draw on roughly 200 underlying studies. The applied criteria they extract are unsurprising — and worth restating because they’re backed by accumulated evidence rather than vibes.

Three things meetings exist to do: solve problems, make decisions, or have a substantive discussion. Sharing routine or non-urgent information isn’t one of them — that’s what writing is for. Agendas aren’t optional; preparation is consistently the strongest single contributor to meeting effectiveness across the underlying studies. And the work of a useful meeting is split across three skills, not one — preparation, in-meeting facilitation, and post-meeting follow-through. Most underperforming meetings fail at one or two of those.

The honest synthesis

What the research as a whole supports. The cognitive cost of interruption is real and measured (Mark et al., 2005). Video meetings have a higher cognitive cost than in-person ones (Bailenson, 2021). Reducing meeting time alone tends to lengthen the workday rather than shorten it (DeFilippis et al., 2022). More sync time does not produce tighter collaboration networks (Yang et al., 2022). Meeting-free days correlate with self-reported gains in productivity, autonomy and satisfaction (Laker et al., 2022). And — backed by survey but consistent with the rest — 71% of senior managers say their meetings are unproductive (Perlow, Hadley & Eun, HBR 2017).

What the research does not support, despite frequent claim. A specific dollar-value or percentage-point productivity gain from cutting any specific number of meetings — the natural experiment says the time gets re-absorbed. The idea that more meetings produce more cohesion — the opposite has been measured. The idea that the trend will resolve itself as remote-work tooling matures — three years of post-2022 data say it has not.

The defensible posture is simple and uncomfortable. Treat sync time as an expensive resource, audit it ruthlessly, replace what writing can replace, and don’t expect the workday to shrink unless you also cut work. The research supports the posture. It doesn’t tell you exactly how much you’ll save. Anyone who pretends otherwise is selling something.