Remote work is almost always written about from a position: defenders cite productivity, critics cite company culture. Almost nobody cites the two randomised experiments that measured both, and that contain a result which is uncomfortable for supporters and critics alike.
This guide starts there, then moves to what the law actually guarantees (more than most workers realise) and finally to the practical rules that follow.
The two experiments worth knowing
The first, 2015. Nicholas Bloom and colleagues published in the Quarterly Journal of Economics a randomised study at a Chinese travel booking company: volunteer employees assigned by lottery to work from home four days out of five, for nine months. Headline result: a performance increase of around 13%, and voluntary resignations cut roughly in half.
The detail almost nobody reports is the breakdown of that 13%: about half came from more minutes actually worked (fewer breaks, fewer sick days, fewer late starts), not from greater efficiency per minute. It’s an honest and ambivalent finding: part of the gain is, literally, more work.
The second, 2024. The same research group published in Nature a randomised experiment on a hybrid model, with two days at home per week. Result: no measurable difference in performance or in performance ratings, and a reduction in quits of around a third.
Put together, the two studies say something quite precise: in a hybrid setup, remote work is neutral on productivity and positive on retention. That’s a far more solid argument than “people work better at home”, and also far less interesting as a headline, which is why it circulates so little.
The cost almost no guide cites
The 2015 study contains a second result, cited much less than the first: remote workers had lower promotion rates, at equal measured performance.
That isn’t a detail. And it’s consistent with another study, published in Nature Human Behaviour in 2022 by Longqi Yang and colleagues, which analysed communication data from tens of thousands of employees before and after a shift to full remote work. Collaboration networks became more static and more siloed: fewer bridges between different groups, more asynchronous communication, fewer new ties.
The practical reading, and the central point of this article:
Remote work doesn’t have a productivity problem. It has a visibility and network problem. Those are two different things requiring two different remedies, and almost nobody implements the second, because it’s less urgent and more decisive.
Everything practical below only makes sense if it accounts for both.
What the law guarantees
Rules vary by country, but two elements are worth checking in your own contract because they’re increasingly common.
In Italy, remote work (“lavoro agile”) is governed by Law 81/2017, which requires:
- A written individual agreement between worker and employer, covering how the work is performed outside company premises, the tools used, and rest periods.
- The agreement must set out the technical and organisational measures ensuring disconnection from technological devices. The right to disconnect isn’t only a principle: it’s a required content of the agreement.
- Equal economic and regulatory treatment compared with colleagues doing the same job on site.
- Insurance protection against accidents and occupational illness remains, including for work performed outside company premises, on the terms set by the legislation.
At EU level, Eurofound tracks how the right to disconnect has been introduced through national law and collective bargaining across member states, and EU-OSHA treats the health and safety risks of remote work as a category in their own right.
Why this is useful: “what does my agreement say about disconnection?” is a technical question grounded in a rule. It works far better than “I’d rather people didn’t message me in the evening”.
The practical rules that follow from the data
| Documented problem | Operational remedy |
|---|---|
| Half the 2015 gain came from more hours worked | A declared, verifiable finishing time, not an intention |
| Networks become siloed | One conversation a week with someone outside your group, in the calendar |
| Fewer promotions at equal performance | Make the work legible: periodic written updates, results with numbers |
| Communication becomes more asynchronous | Write better: context, request and deadline in the same message |
| No boundary between home and work | Physical separation or a closing ritual, repeated daily |
A few points deserve an extra line.
The closing ritual matters more than the desk setup. Shut the laptop, write down tomorrow’s three things, step outside for ten minutes. It signals the end in the absence of a commute doing it for you. It’s mundane and it works.
Meetings are the real cost of remote work, not household distraction. Every meeting that could have been a written message costs everyone in the room, and remotely more get called because they cost the caller less. The most useful rule is to ask for the agenda before accepting: half of them resolve in the reply.
Deep work has to be defended in the calendar, not in your intentions. Two protected ninety-minute blocks beat eight fragmented hours: the method is in time management.
Security is a responsibility that moves with you. Home network, devices shared with family, backups: remotely the company perimeter ends at your house. The minimum measures are in cybersecurity basics.
The visibility remedy, without becoming insufferable
This is the part that generates most resistance, so let me say it plainly: making your work visible isn’t self-promotion, it’s compensating for information that used to travel by itself in an office. In person your manager saw that you were there, that you helped a colleague, that you stayed on the problem. Remotely they see output only.
Three low-cost practices:
- One short written update a week: done, in progress, blocked. Three lines. Solves half the problem.
- Results with numbers, not adjectives, when you report them. It also serves the pay rise conversation and keeps your LinkedIn profile current.
- One conversation a week outside your own group. It’s the direct remedy to the siloing documented by Yang and colleagues, and the one nobody does because it isn’t urgent.
My take: “remote or office” is the wrong question
“Remote or office” strikes me as the wrong question: the data says hybrid is the point where the costs cancel and the benefits remain, and there’s no comparable evidence for either extreme.
The variable that really matters, and appears in no study because it’s hard to measure, is whether your organisation writes things down. Companies that document decisions work well remotely; companies that decide in corridors never will, whatever tools they buy. If you’re in the second kind, the number of days at home is the least relevant problem you have.
Frequently asked questions
Can my employer force me back? Where remote work rests on an individual agreement, as in Italy under Law 81/2017, that agreement governs duration and withdrawal. The concrete conditions depend on it and on the applicable collective bargaining.
Do I have a right to disconnect? In several countries the individual agreement must identify the measures ensuring disconnection. So the right question isn’t “do I have the right”, but “what does my agreement say”. And that’s a question you can put.
Are home costs reimbursed? There’s generally no blanket obligation: it depends on collective bargaining and company agreements, which in several sectors provide allowances or vouchers. Check the applicable contract.
Do people really work more remotely? In the 2015 experiment about half the performance gain came from more minutes worked. That’s a documented risk, not an impression: which is why the finishing time has to be decided rather than left to inertia. If the picture worsens over time, the signals to check are in burnout.
Does remote work damage your career? In the available data the risk concerns visibility and relationship networks rather than performance. It’s a manageable risk, but only if addressed explicitly: written updates and contacts outside your group aren’t courtesies, they’re the remedy.
Sources
- N. Bloom, J. Liang, J. Roberts and Z. J. Ying, Does Working from Home Work? Evidence from a Chinese Experiment, Quarterly Journal of Economics, 2015. Randomised experiment, performance increase of around 13%, resignations halved, lower promotion rates for remote workers.
- N. Bloom, R. Han and J. Liang, Hybrid working from home improves retention without damaging performance, Nature, 2024. Randomised experiment on a hybrid model: no negative effect on performance or ratings, quits down by around a third.
- L. Yang et al., The effects of remote work on collaboration among information workers, Nature Human Behaviour, 2022. More static and siloed collaboration networks after the shift to remote.
- Law 81/2017 (Italy). Rules on remote work: individual agreement, disconnection, equal treatment, insurance protection.
- Eurofound. European data on telework prevalence, work intensity and the right to disconnect.
- EU-OSHA. Health and safety risks connected with remote work.



