Does Hybrid Work Hurt Productivity? What the Evidence Says

The short answer, and why it depends

The best evidence available says that hybrid work, with two days a week at home for office staff doing knowledge work, does not measurably hurt performance and makes people less likely to quit. Fully remote work is a different arrangement, and the evidence on it is mixed: it helped in one call-centre experiment and hurt in a data-entry experiment.

Most debates mix three questions that need separate answers:

1. Which arrangement? Two home days a week is not the same as never coming in.
2. Which work? Routine, individually measured tasks (calls, data entry) behave differently from team-based creative work (software, marketing, finance).
3. Which measure? Output per hour, performance reviews, promotions, quit rates and self-reported productivity can point in different directions.

This article summarizes the randomized trials that underpin most serious discussion of the topic, the Canadian data, and what they cannot tell a ten-person Montreal team. For the policy and news side (federal and corporate return-to-office mandates), see our separate article on return to office in Canada in 2026.

The Nature 2024 trial: design

The most cited study is "Hybrid working from home improves retention without damaging performance" by Nicholas Bloom, Ruobing Han and James Liang, published in Nature on 12 June 2024 ^[1].

All details from the Nature paper. Randomization is what makes this study unusual. Most " remote work productivity" claims compare people who chose to work from home with those who did not, which confuses the effect of location with the kind of people who choose it. Here, a coin-flip equivalent decided who got the hybrid option.

Every employee had an undergraduate degree, and 32% had a postgraduate degree. About two thirds (65%) were men.

What the trial found

Quitting fell by a third. Attrition was 7.20% in the control group and 4.80% in the hybrid group, a one-third reduction ^[1]. The reduction was significant for non-managers, women and employees with longer commutes.

Job satisfaction rose. Mean job-satisfaction scores were 8.19 for hybrid employees and 7.84 for the control group.

Performance did not change. The authors used "null equivalence tests", which go beyond "we found no significant difference" to test whether any effect is small enough to rule out meaningful harm. Hybrid working did not affect performance grades over the next two years of reviews, and there was no evidence of a difference in promotion rates over periods of up to two years.

Code output did not change. For the 653 engineers whose main job was writing code, lines of code submitted per day did not differ between groups.

Why employees valued it. Employees reported that working from home saved commuting time and money and gave them flexibility for occasional personal tasks during the day, which they caught up on in the evening or at weekends.

Volunteering was a signal problem. Before the experiment, non-managers volunteered for hybrid at 35% versus 22% for managers. Women were less likely to volunteer even though they benefited more, which the authors suggest may reflect concern about career signalling. They conclude that full enrolment matters, because volunteering for home days may be read as a lack of ambition ^[1].

Why managers and staff disagreed

Before the trial, managers expected hybrid work to reduce productivity by 2.6% on average; non-managers expected a small gain of 0.7%. After the experiment, the 395 managers in the study revised their view to a perceived positive effect of 1.0% ^[1].

The authors' takeaway is about method rather than ideology: the managers' priors moved once they saw data from their own teams. That is the strongest argument for running a structured trial in your own company rather than adopting someone else's policy.

The earlier call-centre experiment (2015)

Ten years earlier, Bloom and co-authors ran a different experiment at the same company (then called Ctrip). Call-centre employees who volunteered to work from home were randomly assigned to home or office for nine months ^[2].

• Home working led to a 13% performance increase: about 9% from working more minutes per shift (fewer breaks and sick days) and 4% from more calls per minute, attributed to a quieter environment.
• Home workers reported higher work satisfaction and had less turnover.
• But their promotion rate conditional on performance fell.
• When the firm later let everyone choose, over half of the experimental employees switched arrangements, and the gains from working from home almost doubled to 22%, which the authors attribute to learning and selection.

Three differences from the 2024 study matter: this was fully remote (not hybrid), the work was individual and easily measured (calls), and participants were volunteers. The promotion penalty is a warning that visibility still matters even when output rises.

The counter-evidence: fully remote data entry in India

A randomized trial in the Indian data-entry sector points the other way ^[3]:

• Workers randomly assigned to work from home were 18% less productive than those in the office.
• Two thirds of the gap appeared from the first day; the rest came from faster learning in the office.
• Workers who preferred home-based work were 12% faster and more accurate at baseline, so people who choose home are not necessarily the weaker performers.
• Yet those who preferred home were much less productive at home than in the office (27% less, versus 13% less for workers who preferred the office). The authors link this partly to workers with children or home-care responsibilities, who face more demands on their attention at home.

The lesson is not that working from home fails. It is that results from hybrid trials should not be generalized to fully remote work, and that home conditions (space, caregiving, quiet) shape the outcome. A worker in a shared apartment with young children is in a different experiment from one with a spare room.

Comparison of the three trials

Sources: NBER w18871; NBER w31515; Nature, 2024.

What Canadian data shows

Statistics Canada's Labour Force Survey tracks where people actually work. In May 2026 ^[4]:

The hybrid share rose from 6.4% in May 2022 to 10.0% in May 2023 and has varied little since. In other words, fully remote work is shrinking while hybrid has settled at roughly one worker in ten, the arrangement the Nature trial tested.

What Canadians report about their own productivity

Statistics Canada's Research to Insights: Working from home in Canada adds two useful numbers ^[5]:

• About 40% of Canadian jobs can be done from home, with large differences between industries, regions, education and wage levels.
• In February 2021, 90% of new teleworkers (people who usually worked outside the home before the pandemic) reported accomplishing at least as much work per hour at home as in the office.

Statistics Canada itself flags the limits: these are self-reports, whether employers' assessments match them "remains an open question", and productivity effects will likely vary across firms, sectors and arrangements (hybrid versus exclusive telework) ^[5].

Limits of the evidence

Before you quote the Nature trial in a policy memo, know what it does not cover.

• One firm, one country. Trip.com is a large Chinese tech company with structured performance reviews. A Montreal agency of eight people has no comparable review system.
• Graduate staff in creative team jobs. The authors say so themselves: the study focuses on graduate employees in software, marketing, accounting and finance ^[1]. It says nothing about trades, retail or junior roles where learning by watching matters.
• Fixed home days. Everyone in the hybrid group stayed home on the same days (Wednesday and Friday), so office days were full. A "come in when you like" policy is a different treatment.
• Short trial, longer follow-up only for reviews and promotions. Six months of attrition data; two years of reviews and promotions.
• A null result is not proof of zero. Equivalence tests rule out effects larger than a set margin, not every possible small effect.
• Mentoring and innovation. None of the three trials measures how fast juniors learn on team work or how many new ideas a team produces. The India study shows learning can be faster in the office for routine work.

Turning evidence into a policy for a small team

A practical reading of the evidence for a Montreal team of five to twenty:

1. Prefer hybrid over fully remote if you are choosing between them. The strongest evidence of "no harm plus lower quitting" is for hybrid.
2. Fix the anchor days. The trial used set days. Choose two or three days when everyone is in, and put meetings, reviews and onboarding on those days.
3. Enrol everyone by default. The trial's lesson on signalling: do not make hybrid an opt-in that people fear asking for.
4. Protect juniors. Ask new hires to spend more days in during their first months, and pair them with a senior on anchor days.
5. Watch home conditions. If someone's home setup is poor (shared space, caregiving at home), offer an alternative such as a coworking desk near home.
6. Write it down. In Quebec, a written policy sets expectations on schedules, equipment and costs; our guide to a hybrid work policy for small Quebec teams covers the content.

A simple trial design for your team

You will not get a randomized trial with 1,612 people, but you can borrow its structure.

Design (90 days)

Metrics that are hard to game

• Output you already count: tickets closed, deliverables shipped, billable hours, sales calls, client response time.
• Quality: rework, client complaints, bugs reopened.
• Retention signals: resignations, sick days, a short monthly satisfaction pulse (one to ten).
• Collaboration: time to first response from a colleague; number of meetings that had to be rescheduled.

Traps to avoid

• Hours online as a measure. Activity tracking measures presence, not output.
• Changing two things at once. Do not launch a new tool or reorganize during the trial.
• Ignoring seasonality. Compare like with like: December is not a typical month.
• Asking only managers. In the Nature trial, managers and staff started with opposite views; survey both before and after.
• Small numbers. With eight people, one resignation is not a trend. Treat the trial as structured feedback, not proof.

Anchor days need a room where the whole team fits. If your office is small, our guide to meeting room hours for hybrid teams in Montreal shows how to size them.

A team testing hybrid in Montreal can book 2727 Coworking's day passes and desks in Griffintown for anchor days without signing a lease first.