Why a simple question gets a messy answer
“Are people more productive working from home?” sounds like a question with a number attached. In practice, the answer depends almost entirely on what is being counted, who is counting, and which workers are included. Two reports can both be honest and still point in opposite directions, because they are measuring different things under the same word.
This article is about that gap. It does not try to settle whether remote work helps or hurts output. It explains why the evidence is hard to compare and how to read a productivity claim without being misled by it.
What “productivity” means in different settings
In economics, productivity usually means output divided by input. A factory can count units made per hour of labour. A national statistics office typically measures output per hour worked across the whole economy, or in a sector. These are real and useful figures, but they are very coarse. They mix together changes in the kinds of jobs that exist, the tools in use, and the hours people report.
In a company, productivity often means something more specific: tickets closed, calls handled, lines of code merged, sales made, documents processed. Those measures work reasonably well for repetitive tasks where each unit is similar. They work badly where the valuable part of the job is judgement, coordination or creative problem-solving, because the countable thing is only a small part of what the person contributes.
Different productivity measures also answer different questions:
- Output per hour asks how efficiently time is used.
- Output per worker asks how much each person produces, regardless of hours.
- Quality-adjusted output asks whether the work was any good.
- Output per unit of total cost asks whether the arrangement pays, including office space, travel and equipment.
A remote arrangement might look better on one of these and worse on another, which is one reason summary headlines are unreliable.
Why knowledge work is especially hard to count
Much of the remote work debate concerns office-based knowledge work: analysis, writing, management, design, planning. Several problems recur.
Output is not standardised. One report is not interchangeable with another. Counting them tells you little about their value.
Work is collaborative. A person’s contribution may show up as someone else’s faster progress, which individual metrics rarely capture.
Effects are delayed. Mentoring, relationship-building and the flow of informal knowledge may matter for a team’s long-term performance but rarely appear in this quarter’s figures.
Managers’ judgement is subjective. Supervisor ratings are common, but they can be influenced by what is easy to see. Whether visibility affects ratings is itself debated.
How the data is typically gathered
Evidence on remote work and productivity tends to come from a handful of sources, each with strengths and blind spots.
Surveys of workers. These are cheap and broad, but they rely on self-reported behaviour: people estimating their own output, focus or hours. Estimates can be shaped by mood, by what people think the employer or researcher wants to hear, and by how the question is asked.
Employer or administrative records. Metrics such as calls answered or tasks completed are more objective, but they usually cover only roles where work is easy to count, so they may not describe knowledge work in general.
Experiments and pilots. When some employees are assigned to work remotely and others are not, comparisons are cleaner. But experiments are often run in specific firms and roles, and results may not carry over to other settings.
National statistics. These capture the whole economy but cannot separate remote work from everything else changing at the same time, such as technology, the business cycle or shifts in which industries are growing.
Common traps when reading the numbers
Who is in the sample. People who choose remote arrangements may differ from those who do not, in role, experience or motivation. Selection effects of this kind can make a group look better or worse for reasons unrelated to location.
Confounding changes. Remote work expanded alongside new software, changed hiring and altered business conditions. Any of these can shift output, which is why confounding variables make it hard to credit or blame one factor.
Hours versus output. Some studies find people report working longer days at home. If hours rise and output rises less, per-hour productivity can fall even when total output is higher. If a study counts only one of these, the story changes.
Time saved on commuting. Reclaimed travel time may be reinvested in work, rest or family. Whether that counts as “productivity” is a definitional choice, and commuting patterns vary widely between places, so the effect differs from one workforce to the next.
The comparison point. A remote worker may be compared with an office worker in the same firm, with the same person before the change, or with a national average. Each gives a different answer.
Different definitions, different stories
Suppose a company reports that remote employees closed more tasks. A reader might reasonably ask whether the tasks were comparable, whether quality was checked, whether people worked longer hours, and whether the most capable staff were the ones allowed to work remotely. Suppose a separate survey says workers felt less productive at home. That might reflect real friction, such as weaker communication, or it might reflect a lack of the visible cues that signal busyness in an office. Both findings can be accurate and still not contradict each other.
The same applies to arrangements that mix locations. Studies of hybrid work often rely on what employers choose to report, and they raise their own questions about bias. Similar caution applies to experiments with a four-day week, where shortening the week changes both hours and how work is organised.
A practical way to read a productivity claim
- Ask what was actually measured, and whether it is output, hours, quality or opinion.
- Ask who was studied and how they ended up in that group.
- Ask what the comparison is, and whether other things changed in the same period.
- Be wary of a single figure presented as the answer for all jobs.
- Treat confident forecasts about the future of work as arguments rather than established findings, since the evidence is still developing and tends to be context-specific.
Productivity is better thought of as a family of measures than a single quantity. A trend story that names its measure, its sample and its comparison group deserves more trust than one that offers only a conclusion, and it is reasonable for several rival explanations to remain open until better evidence arrives.