Managing by Henry Mintzberg: What the Evidence Says

Every claim here is checked against the primary research. How we check.

TL;DR — Mintzberg’s «Managing» (2009) says the textbook manager who plans, organizes, coordinates and controls does not exist, and that the real job is fast, fragmented and spoken. The observational core of that claim is 50 years old and mostly replicates. The book’s own frameworks have never been tested. And Mintzberg’s dismissal of formal management training is the part the evidence actually contradicts.

The original data: five chief executives, one week each, 890 pieces of mail and 368 verbal contacts. Half of all activities lasted under nine minutes. Only 10% ran over an hour.
Replications back the brevity finding. A 1983 structured-observation study of four top managers reported that Mintzberg was supported «in all important dimensions».
The largest modern dataset — 1,114 CEOs in six countries, 42,233 logged activities — finds something Mintzberg’s five-manager sample could not: only 4% of CEO time went to emergencies, and under 10% of planned activities were cancelled. CEO days are mostly planned.
That same dataset adds what Mintzberg never had: a one-standard-deviation shift in CEO behavior tracks a 7% difference in sales. How you allocate the hours predicts firm output.
The killer: Mintzberg argues management cannot be taught in a classroom. A randomized trial in Indian textile plants raised productivity 17% in one year by teaching standardized management practices. Both things can be true. This page explains how.

Verdict

Read the notes, not the book. Confidence: moderate-to-high on the descriptive claims, high on the criticism.

«Managing» is 300 pages restating a finding Mintzberg published in twelve pages in 1975, plus a set of typologies — three planes, twelve postures, thirteen conundrums — that no one has ever tested. The descriptive core is real and holds up better than most popular management claims. The prescriptive core is unverified, and the one prescription that has been tested points the other way.

The claim on trial

Mintzberg’s central falsifiable claim: managerial work is not planning, organizing, coordinating and controlling. It is a brief, varied, discontinuous, interruption-driven, overwhelmingly oral practice — and this pattern is stable across hierarchical levels, sectors and decades.

This is testable in a way almost no popular management book’s thesis is, because Mintzberg made it by watching people with a stopwatch rather than by asking them.

What the original observation actually showed

The evidence base is Mintzberg, Management Science, 1971: five managers, structured observation, one intensive week each. Organizations spanned a consulting firm, a technology company, a hospital, a consumer goods company and a school system. The paper’s own summary says the manager works «at an unrelenting pace, work that is characterized by variety, discontinuity and brevity».

The numbers appeared in «The Manager’s Job: Folklore and Fact», Harvard Business Review, first published 1975 and reprinted in 1990. Mintzberg logged 890 pieces of incoming and outgoing mail and 368 verbal contacts. Half of the five executives’ activities lasted under nine minutes; only 10% exceeded an hour. 93% of verbal contacts were arranged ad hoc. Exactly one of 368 contacts was unrelated to a specific issue and could be called general planning. 78% of their time went to oral communication. One executive came in on a Saturday to clear 142 pieces of mail «to get rid of all the stuff».

That is a genuinely small sample carrying an enormous literature. It is also, unusually, real data. Most of what gets sold as management insight is a consultant’s anecdote.

«Managing» itself adds 29 days of observation — one day with each of 29 managers, per Mintzberg’s own description of the book. Twenty-nine single days is a thinner base than the 1971 study’s five full weeks, and the book leans on it to build much larger structures.

Does the fragmentation finding replicate

Mostly, with one large qualification.

Kurke and Aldrich, Management Science, 1983 ran structured observation on four top managers for one week each and reported that Mintzberg’s field study «was supported by our replication in all important dimensions». Martinko and Gardner, Journal of Management Studies, 1990 replicated again and synthesized the earlier work, concluding that the results «generally support earlier conclusions regarding the brief, varied, fragmented and interpersonal nature of managerial work». They also found managerial behavior varied with environment and demographics but not with performance — worth holding onto.

Thirty years on, Tengblad, Journal of Management Studies, 2006 observed Swedish top executives using Mintzberg’s method and got a partly different picture: a much larger workload, more time with subordinates in groups, less administrative work — and, in his words, «fragmentation of time — in previous studies highlighted as a central tenet of managerial work — was not as prevalent in the new study».

Then the sample size problem gets solved. Bandiera, Prat, Hansen and Sadun, Journal of Political Economy, 2020 collected diary data on 1,114 CEOs of manufacturing firms in Brazil, France, Germany, India, the UK and the US: 42,233 activities, 225,721 fifteen-minute blocks, an average of 50 working hours per CEO week. The average CEO spent 70% of time interacting with others. That confirms Mintzberg’s «managing is talking» finding at 200 times the scale.

It does not confirm the interruption-driven part. The researchers asked, activity by activity, whether it was a response to an emergency. Only 4% of CEO time was. They compared each morning’s planned schedule against each evening’s actual one: just under 10% of planned activities were cancelled. Their conclusion is that CEO activities «largely reflect conscious planning vs. mere reactions to external contingencies».

That is the sharpest available correction to Mintzberg. Managerial days are short-segmented, but they are not chaotic. The segments are largely the ones the manager put on the calendar. «Fragmented» and «reactive» are different claims, and the second one is weaker than «Managing» says.

For a US-specific reading, Porter and Nohria’s CEO time-use study, Harvard Business Review, 2018 tracked 27 large-company CEOs 24/7 for 13 weeks, about 60,000 hours of data. They report a 62.5-hour average week with 61% of work time in face-to-face interaction. The medium changed since 1971. The oral dominance did not.

Why fragmentation costs something

Mintzberg describes the fragmentation and largely accepts it. The interruption literature says what it costs, and the answer is more interesting than «interruptions ruin your day».

Mark, González and Harris, CHI, 2005 shadowed 24 information workers — 7 managers, 9 analysts, 8 developers — for over 700 formal hours. People averaged 11 minutes 4 seconds in a working sphere before switching. 57.1% of working-sphere segments were interrupted. 77.2% of interrupted work was resumed the same day, but resumption took an average of 25 minutes 26 seconds, with multiple intervening activities in between.

The lab result is the surprise. Mark, Gudith and Klocke, CHI, 2008 interrupted 48 participants during an email task and found they completed interrupted work faster, with no quality difference. The cost showed up elsewhere: significantly higher stress, frustration, time pressure and effort. Their summary is that people compensate for interruptions by working faster, «but this comes at a price».

Leroy, Organizational Behavior and Human Decision Processes, 2009 named the mechanism: attention residue, the part of the previous task still occupying you after you switch. That is the honest version of the fragmentation cost. Not lost output. Lost slack.

Where the book is wrong: management is teachable

Mintzberg’s most forceful prescription is that management is a practice learned through experience, not a profession taught in a classroom, and that MBA-first training produces the wrong managers. He has been arguing this since «Managers Not MBAs» and repeats it in «Managing».

The field-experimental evidence goes the other way, hard.

Bloom, Eifert, Mahajan, McKenzie and Roberts, Quarterly Journal of Economics, 2013 ran a randomized management intervention across 28 plants operated by 17 large Indian textile firms. Treatment plants received free consulting on a checklist of standard practices — quality defect logging, inventory control, preventive maintenance, production tracking. Control plants got diagnostic work only. Adopting these practices raised productivity by 17% in the first year, through better quality, higher efficiency and lower inventory, and within three years the treated firms opened more plants. The reason the practices had not been adopted already was mostly informational: the owners did not know they existed.

These are not gains that evaporate. Bloom, Mahajan, McKenzie and Roberts, American Economic Journal: Applied Economics, 2020 returned nine years later. About half the adopted practices had been dropped, but a large, significant gap between treatment and control plants remained. Notably, the two most-cited reasons for the drop were managerial turnover and lack of director time — Mintzberg’s own scarce resource, working against durable practice.

The observational evidence points the same way. Bloom and Van Reenen, Quarterly Journal of Economics, 2007 scored 732 medium-sized manufacturing firms in the US, France, Germany and the UK and found management scores strongly associated with productivity, profitability, Tobin’s Q, sales growth and survival. Bloom, Brynjolfsson, Foster, Jarmin, Patnaik, Saporta-Eksten and Van Reenen, American Economic Review, 2019 scaled this with the US Census Bureau across two waves of 35,000 manufacturing plants: management practices account for more than 20% of the variation in productivity, comparable to or greater than R&D, ICT or human capital.

Even the MBA question has been probed directly. Bloom, Lemos, Sadun and Van Reenen, Review of Economics and Statistics, 2020 scored management in hospitals across nine countries and found that hospitals near universities offering both medical and business education had better management practices, more MBA-trained managers, and lower heart-attack mortality — relative to hospitals near universities offering only one or neither.

So how do both survive? Cleanly, if you separate two things Mintzberg conflates. The content of good management — measure defects, track output, set targets, promote on performance — is codifiable, teachable and raises output. The execution of a manager’s day — reading a room, sequencing a difficult conversation, deciding what not to attend — is a craft learned by doing. Mintzberg observed executives and concluded management is not a science. What he actually observed is that the delivery is not a science. The checklist underneath it demonstrably is. «Managing» spends 300 pages on the delivery and treats the checklist as bureaucratic noise.

The frameworks nobody tested

The 1971 ten roles — figurehead, leader, liaison, monitor, disseminator, spokesperson, entrepreneur, disturbance handler, resource allocator, negotiator — are the most-taught part of Mintzberg’s work and among the least validated. Pearson and Chatterjee, Journal of Management Development, 2003 tested the ten roles in a four-country Asian sample and found that «although the roles overlap considerably, they are acted out in a very different manner». Overlapping categories are a construct-validity problem, not a footnote.

The frameworks new to «Managing» — the three planes of information, people and action, the twelve managerial postures, the thirteen conundrums — fare worse. Searching Crossref on 10 August 2026 for empirical validations of the «Managing» model returned no study testing it: no factor analysis of the three planes, no measurement instrument for the postures, no test of whether the thirteen conundrums predict anything. Untested is not the same as refuted. But a framework published in 2009 that nobody has attempted to measure in seventeen years should be read as vocabulary, not as findings. Every claim in this section is why we run these book reviews against primary sources rather than against jacket copy.

Who should actually read it

New managers who feel like frauds because their day is nothing like the textbook. The book will tell you the textbook was wrong. That is a genuinely useful thing to learn, and it costs you twelve pages, not three hundred.
Researchers on managerial work. The 1971 and 1975 papers are the field’s founding documents and are worth reading in the original.
Anyone hoping for a management system. Skip it. The book is diagnostic by design and says so.
Anyone deciding whether to fund management training. Read the Indian textile trial instead. It has a control group.

One thing to try

Do Mintzberg’s experiment on yourself, then check it against the modern finding. For one week, log every work activity longer than 15 minutes — that is the exact unit used in the 1,114-CEO study. At the start of each day write your planned activities; at the end mark which happened, which were cancelled, and which were genuine emergencies.

Then compare. The CEO benchmark is 4% emergency time and under 10% cancellation. If your numbers are much worse, the problem is not the nature of managerial work — it is your calendar, and it is fixable. If your numbers are close, stop apologizing for a fragmented day. It is the job.

Get the book

Find «Managing» on Amazon — as an Amazon Associate, The Boring Work earns from qualifying purchases (disclosure).

The boring bottom line

Mintzberg won the descriptive argument and lost the prescriptive one. Managers do work in short oral bursts; five decades of structured observation and a 1,114-CEO diary study agree. But those days are more planned than «Managing» admits, the book’s own frameworks have never been measured, and the strongest field experiment in the literature says a boring checklist of teachable practices raised plant productivity 17% in a year. Read the 1975 article. It is free, it is twelve pages, and it carries the part that survived.

Managing by Mintzberg: The Fragmented Manager, Re-Checked