Productivity in the 2020s: What Changed

By Victor Hoang · Updated Oct 24, 2026

When offices emptied in 2020, nobody had a new way to measure work. The tools were the ones built for factories and national statistics: hours, output, attendance.

The 2020s changed productivity in one specific way. For the first time at scale, the place where work happens stopped being a given. Organizations had to decide what they were actually measuring. The decade began in the second quarter of 2020, when official US productivity statistics recorded the largest quarterly jump since 1971. It then moved through remote and hybrid work, a trust crisis between leaders and staff, four-day-week trials, office mandates and generative AI. The throughline is that the yardstick (output per hour) is more than a century old, while the way people work was rearranged in a few years. Every shock since 2020 has been an argument about what to count.

The 2020s are the story of those tools meeting a workplace they were never built for.

Timeline at a Glance

The Pig-Iron Yardstick (1899-1960s)

Output per hour becomes a number a government can track

The precursor of every productivity argument in the 2020s is the idea that work can be counted per person per hour. It was not new in 1911, but Frederick Winslow Taylor made it a management method. In The Principles of Scientific Management, he described the Bethlehem Steel pig-iron gang: about 12.5 long tons per man per day before his changes, and a target of 47 to 48 tons after them, with the worker he calls Schmidt paid about 60 percent more. The principle was that a manager studies a task, sets a standard and pays for hitting it.

Governments adopted the same lens. BLS lists a 1899 release called Hand and Machine Labor covering 60 manufacturing industries as early productivity work. In 1940 it created a productivity division, and a statute directs the bureau to make continuing studies of productivity and labor costs. In 1959 it published labor productivity for the total private economy back to 1909, which is still the backbone of the statistic quoted in headlines. The fuller story of how the number was built is in our history of productivity measurement, and the factory logic behind it is in the history of time and motion studies.

To be precise about the lineage: Taylor and the BLS built the yardstick for physical output. They were precursors, not an early version of the 2020s debate, because the work they measured could be seen, weighed and counted. The new problem, which would take decades to bite, was that most work stopped being countable that way. That shift is traced in the history of knowledge work.

What it means for teams today: every dashboard of "hours" or "tasks" descends from this era. It works when output is tangible and misleads when it isn't.

The Missing Gains (1970s-2019)

Computers arrive, but the statistics refuse to show it

The second turning point was the opposite of a boom. BLS data show that productivity in the first quarter of 1971 rose 12.3%, the previous time a quarter had moved that far. In 1987 the economist Robert Solow wrote that "You can see the computer age everywhere but in the productivity statistics." The quote, from Solow in 1987, became the Solow paradox, and Jack Triplett's 1999 Brookings commentary lists several explanations, including that the effects of new technology may only be visible with a long lag, a comparison he draws with the spread of electricity.

BLS meanwhile refined the measure. It began publishing multifactor productivity in 1983, and renamed it total factor productivity in 2021 as a terminology change only. The tool that mattered most for the 2020s was still the simple one: output compared with hours worked. In the long business cycle from the fourth quarter of 2007 through the fourth quarter of 2019, nonfarm business productivity grew at an annualized rate of 1.5 percent. That is the baseline against which the 2020s are read.

The same period built the habit of measuring knowledge work by presence and activity. Offices built around presence, as the history of the office and the history of the open-plan office show, are part of that story, and the productivity paradox article tells the Solow story in full.

What it means for teams today: a gain from new tools can take years to appear in results. Judge adoption on a long enough horizon before declaring it a failure.

The Sudden Jump (2020-2021)

A statistical record that mostly shows who stopped working

In the second quarter of 2020, the BLS revised its estimate of nonfarm business productivity to an increase of 10.6 percent at an annual rate, the largest since the 12.3% of the first quarter of 1971. Behind it, output fell 36.8 percent and hours worked fell 42.9 percent. Productivity went up because hours fell faster than output, and that arithmetic is the first lesson of the decade: a rising ratio is not the same as people getting better at their jobs.

At the same time the place of work changed. BLS reported that the share of employed people who worked at home on days they worked rose from 22 percent in 2019 to 42 percent in 2020, on a comparison covering May 10 to December 31. Managers who had judged by presence now had to judge by something else. The full story of how that shift came together is in our history of remote work, and the tools that carried it are covered in the history of collaboration tools.

The practical guidance that grew out of it sits in our articles on remote work productivity and async communication.

What it means for teams today: treat any one-quarter productivity figure with suspicion. Ask what happened to the numerator and the denominator separately.

Productivity Paranoia (2021-2023)

Staff say they are productive, leaders say they can't tell

By 2022 a new problem had replaced the logistics of working from home. On September 28, 2022, Microsoft published results from a survey of 20,000 people in 11 countries. It reported that "87% of employees report they are productive at work", while 85% of leaders said the shift to hybrid work had made it challenging to have confidence that people are being productive. The company named the gap "productivity paranoia."

The gap exposed what the old yardstick could not see. Activity was easy to count (meetings, messages, logged hours) and results were not. The strain that constant measurement and pressure can put on people is covered in the history of workplace wellbeing and burnout, and the clock-based tradition behind it is in the history of the time clock.

It also made an old distinction practical: busy is not the same as productive. Teams that had run on observation had to define what good output looked like in writing, for roles where it had never been stated.

What it means for teams today: if your managers say they can't tell whether people are productive, the fix is a clearer definition of output, not more monitoring.

The Evidence Years (2022-2024)

Four-day weeks and hybrid weeks get tested, not just argued

This era overlaps with the last one. The loudest debate of the early 2020s shifted from opinion to experiment. In the UK pilot, which ran from June to December 2022, 61 companies and around 2,900 workers took part in what was then described as the world's largest four-day week trial. By the end, 56 of the 61 companies were continuing and 18 had made the change permanent. The organizers reported that 71% of employees had lower burnout, resignations fell by 57%, and revenue stayed broadly the same, rising 1.4% on average. Those are the pilot's own published results, from a self-selected group of employers. The longer story is in the history of the four-day week.

Hybrid work got a randomized test too. Nicholas Bloom, Ruobing Han and James Liang ran a six-month trial of 1,612 employees at Trip.com from August 2021 to January 2022, published in Nature in June 2024. With two days a week at home, quit rates fell by about one-third, from 7.20% to 4.80%, and there was no significant effect on performance reviews or on promotions over two years. Managers who had expected hybrid to cut productivity by 2.6% afterward expected a 1.0% gain.

By then the long-run level had settled. Barrero, Bloom and Davis describe working from home as having stabilized at about 25% of all paid workdays as of 2025.

What it means for teams today: location and schedule are testable. Pick a measure you trust (retention, delivery, quality), run a defined trial and read the result.

Mandates and Machines (2024-present)

Offices recall workers while AI assistants arrive at the desk

Two forces now run in parallel. The first is pressure to return. On January 20, 2025, a presidential memorandum directed agencies to "terminate remote work arrangements" and require full-time in-person work at duty stations as soon as practicable, subject to exemptions agency heads decide. Even so, BLS reported that in 2025, 35 percent of employed people did some or all of their work at home on days they worked.

The second is generative AI. The first large field study, Generative AI at Work by Erik Brynjolfsson, Danielle Li and Lindsey Raymond, covered 5,179 customer support agents. It found about a 14% average increase in issues resolved per hour, with novice and low-skilled workers improving by about 34% and experienced workers seeing minimal impact. A Danish study of 25,000 workers in 11 occupations, released in May 2025, found modest time savings of about 3 percent on average and no significant effect on earnings or recorded hours. The two results do not contradict each other: one measures a defined task in a single firm, the other the labor market across occupations. The tools themselves have a history, covered in the history of AI productivity tools.

The statistics keep a running score. BLS's September 3, 2026 release shows that since the fourth quarter of 2019, nonfarm business productivity has grown at an annualized rate of 2.1 percent, against 1.5 percent in the prior cycle, with the second quarter of 2026 at 1.4 percent. Whether that is the hybrid era, AI, or something else is not settled by the data.

What it means for teams today: neither a mandate nor a tool answers the question of what good output is. Measure the work you want, then choose the location and tools that help.

Key People

Key Facts

Key Facts: Productivity in the 2020s

• BLS revised second-quarter 2020 productivity to +10.6%, while hours fell 42.9%.
• The share working at home on days worked rose from 22 to 42 percent between 2019 and 2020.
• Microsoft's 2022 survey found 87% of employees said they were productive while 85% of leaders lacked confidence that people were.
• In the UK pilot, 56 of 61 companies kept the four-day week afterward.
• A randomized hybrid trial found quit rates fell by about one-third.
• Since late 2019 productivity has grown at 2.1 percent a year, against 1.5 percent from 2007 to 2019.

The Pattern Behind the History

We call this the Ratio-Shock Pattern. It is our own synthesis of this timeline, not a published model.

1. The yardstick outlives the work. Output per hour dates to Taylor and the BLS, yet it is applied to work they never imagined.
2. A shock moves the ratio before it moves the work. The 1971 and 2020 spikes came from hours collapsing, not people improving.
3. Unmeasured work breeds suspicion. Productivity paranoia grew where output was not defined, so activity stood in for it.
4. Experiments replace arguments. Four-day and hybrid trials turned opinion into evidence.
5. Each new tool restarts the loop. AI is being judged on task studies and labor-market studies that disagree.

What It Teaches Teams Today

• Define output in writing. Say what a good week looks like for each role.
• Read ratios with both parts. A productivity figure needs its output and its hours.
• Run trials. Test schedules and tools against a measure you chose in advance.
• Separate tasks from jobs. An AI gain on one task is not an across-the-board gain.
• Expect reversals. Mandates and exemptions swing, so keep decisions tied to evidence.
• Place this in context. The long arc is in our history of productivity, with the dated sequence in the productivity timeline.