Remote, hybrid or office: which way of working fits your team?

The office debate can become a contest between two photographs: a productive person at a kitchen table and an empty desk someone is still paying for. Neither photograph tells you whether the customer got a good answer, a newcomer learned anything, or the team made a useful decision.

I'm Daniil Shakhovskiy. My starting point is the work, the people doing it, and the dependencies between them. A location policy should explain how those parts will function together. The research gives us useful trade-offs to investigate, rather than one winning arrangement for every company.

What is the difference between remote, hybrid and office work?

Remote work happens away from a usual shared workplace. Working from home is one version of it. Hybrid combines remote and office time. Office-based work puts people in a shared workplace, although their actual collaborators might still be elsewhere.

The schedule matters. Two home days, four home days, and a distributed team with occasional gatherings are different arrangements. So are coordinated office days and a policy where everyone chooses different days. Count the useful overlap, not just the chairs.

What does research say about productivity and retention?

Can two home days work well?

A 2024 Nature study at Trip.com randomised 1,612 university-educated employees in two Shanghai divisions to an offer of two home days weekly or five office days. Over the six-month experiment, quit rates fell from 7.2% to 4.8%: a 2.4 percentage-point reduction, or 33% relatively. Performance-grade and coding-output tests supported no material difference within specified bounds.

That is encouraging evidence for this hybrid policy in this company. It is not a fully remote experiment. Interns and probationary newcomers were excluded. Performance follow-up continued, but hybrid was rolled out after the experiment, so this was not two years of continuously different randomised schedules.

Can home working increase individual output?

The earlier Ctrip call-centre experiment involved 249 eligible volunteers over nine months. The home-working group spent four days at home and one in the office. Measured performance improved by about 13%, largely through more minutes worked per shift, with a smaller contribution from calls per minute.

Participants had suitable home conditions, and the work was measurable call-centre activity. Promotion probability conditional on performance fell. The result deserves attention, but neither the output gain nor the promotion finding should become a universal forecast for every employee or role.

What can happen to collaboration and learning?

A study of 61,182 Microsoft employees examined the initial 2020 shift to remote work using a natural-experiment design. It estimated that the share of collaboration time involving cross-group connections fell by about 25% relative to its earlier level. Networks became more siloed. This was a communication-network result, not a 25% productivity decline, and the design depended on assumptions about the pandemic transition.

In The Power of Proximity to Coworkers, published in 2026, researchers used office closures and return mandates at one online retailer to estimate effects of working near teammates. Proximity increased coding feedback by 18.3%, with benefits concentrated among newer and younger engineers, while experienced engineers produced less own code.

This illustrates a potential trade-off between individual output and helping others learn. It does not establish a five-day office rule for juniors or evaluate every form of remote onboarding. Being near the relevant colleague matters more to that argument than entering a building somewhere.

These studies examine different jobs, schedules, outcomes, and periods. Averaging their percentages into a universal remote-work score would produce a tidy number with very little meaning.

Which working model fits which team?

I would use the following as a starting discussion, then test the assumptions locally. A company may need several arrangements because a role often contains both physical and digital tasks.

Customer requirements should be specific. Does someone need to be physically present, available at a particular time, or able to access a controlled system? Those are different constraints. Separate them before writing an entire company policy. For colleagues whose work must stay on site, discuss useful flexibility in scheduling, handovers, and administrative tasks rather than treating home working as the only form of autonomy.

How should hiring and onboarding influence the choice?

Remote hiring can let you consider people beyond commuting distance, where you can practically employ and support them. It also makes time zones, communication habits, equipment, and access to colleagues part of the job design. Explain those expectations before someone joins.

For newcomers, name the people who will teach them and protect those people's time. Arrange examples, pairing, introductions, and a route for questions. If learning requires being together, coordinate the relevant people rather than sending a new hire to sit alone near a coffee machine.

Check whether feedback is timely and whether the newcomer can increasingly handle real work. A mentor's contribution can be missed if you measure only their own output. This connects directly to the people who help a business grow: teaching and connecting colleagues deserve time and recognition.

What do autonomy, belonging and psychological safety have to do with location?

Autonomy means having meaningful influence over how work is done. Belonging involves feeling included and able to access other people. Neither follows automatically from a home address or an office badge.

Wang and colleagues' work-design research combined 39 interviews with a separate survey of 522 employees during emergency home working in China. Social support was associated with fewer reported difficulties; monitoring and workload were associated with more work-home interference. These were associations in a particular context, not proof that remote work causes loneliness or that autonomy fixes every problem.

I would ask whether people can focus, get help, take part, and stop working. Home is not equally suitable for everyone. Let people discuss practical constraints without requiring private disclosures. Agree on response expectations and avoid turning flexibility into an unspoken requirement to be available forever.

Psychological safety concerns whether interpersonal risks, such as asking a basic question or reporting a mistake, feel safe. Amy Edmondson's study of 51 work teams found an association with learning behaviour. It was organisational research in a manufacturing company, not an experiment comparing remote and office work.

The practical question is how managers respond when someone says, “I do not understand,” or, “This plan is going wrong.” Provide spoken and written ways to contribute. Explain decisions outside the room where they happened. Make space for disagreement without making accountability optional. An office cannot do this on a manager's behalf.

How much does remote or hybrid work actually cost?

Here is a deliberately hypothetical comparison, not a forecast. Assume 12 people, a four-week month, and a loaded labour rate of €50 per hour. Monthly workspace costs are €9,000 for office, €6,000 for hybrid, and €0 for remote. Monthly support allowances per person are €0, €60, and €90 respectively.

Monthly cash cost = workspace cost + team size × support per person. That gives €9,000, €6,720, and €1,080. These inputs are illustrative, not market prices or a complete company budget. A remote team may still pay for gatherings, equipment, or workspace.

Assume the same two hours of coordination friction per person per week in every model. Here, friction means avoidable waiting, repeated explanations, or preventable rework, rather than all collaboration. Twelve people × two hours × four weeks gives 96 team hours, valued at €4,800. This is a capacity proxy for time, not an additional invoice. Do not add it to salary expenditure and count the same labour twice.

Commuting totals are separate personal time: people × office days × four weeks × one hour. They are not automatically company savings or working hours.

Hybrid saves €2,280 in assumed monthly cash costs against office. The calculation €2,280 ÷ (12 × 4 × €50) gives 0.95 additional friction hours per person per week with an equivalent assigned value. That is a way to discuss trade-offs, not a claim that cash savings disappear, a net-profit calculation, or a productivity prediction.

Fixed leases often do not shrink when attendance does. Test realistic inputs, keep cash and capacity separate, and remember that service quality, learning, and employee experience do not fit neatly into one hourly rate. The studies' 13% output gain and 33% relative attrition reduction are not constants in this model.

Compare your team’s work models ↗

TRY YOUR OWN NUMBERS

A four-week planning model. All starting values are illustrative, including equal friction across formats. Change them to match your team.

People in the teamLoaded hourly cost (€)Round-trip commute (minutes)

Office

Office days / person / weekWorkspace / month (€)Home-work support / person / month (€)Avoidable friction / person / week (hours)

Hybrid

Office days / person / weekWorkspace / month (€)Home-work support / person / month (€)Avoidable friction / person / week (hours)

Remote

Office days / person / weekWorkspace / month (€)Home-work support / person / month (€)Avoidable friction / person / week (hours)

Workspace means the actual bill you expect to pay. Two office days do not automatically reduce a five-day lease. Friction means avoidable waiting, interruptions or duplicated work, not all meetings or mentoring.

Office

Operating cash / month

€9,000

Value of avoidable paid time

€4,800

Team commuting hours / month

240 h

Hybrid

Operating cash / month

€6,720

Value of avoidable paid time

€4,800

Team commuting hours / month

96 h

Remote

Operating cash / month

€1,080

Value of avoidable paid time

€4,800

Team commuting hours / month

0 h

Hybrid saves €2,280 in monthly operating cash versus the office example. That equals 0.95 extra hours of friction per person per week at the stated hourly cost.

Paid-time value is a capacity estimate, not an extra cash invoice. Salary is already a budgeted cost. Commute hours are reported separately as personal time. This model does not price retention, learning, quality or wellbeing, and it cannot choose a work policy for you.

Reset example

How can you test a working arrangement fairly?

Start with a baseline and a clear question. Choose a period long enough to cover the team's work cycle, with a review date and agreed reasons to adapt. A short pilot cannot settle long-term retention or development.

• Useful output: completed work, quality, rework, and customer outcomes appropriate to the role.
• Coordination: blocked handoffs, decision delays, and access to other teams.
• Learning: feedback access, mentoring time, and newcomer progress.
• Experience: whether people can focus, get support, contribute, and disconnect.
• Resources: actual cash changes and time demands, recorded separately.

Inspect differences by role and tenure. Keep track of other changes, such as staffing or workload, that could explain a result. A simple before-and-after comparison is useful feedback, but weak proof that location caused the change. Avoid using keystrokes, badge counts, or online status as substitutes for useful work.

The same discipline applies to introducing AI into a business: define the problem, involve the people, test a manageable change, and review what actually happened.

I would choose the arrangement that gives this team a credible way to serve customers, learn, and work together. Then I would keep listening. The policy should be clear enough to use and flexible enough to improve when the evidence changes.

Questions people ask

Is hybrid work always better than fully remote or office work?

No arrangement is a universal winner. Studies examine particular jobs, schedules, and outcomes, and a successful two-day home-working policy is not evidence for every remote setup. Start with physical tasks, customer needs, team dependencies, and learning requirements. Then make collaboration and support deliberate, test the arrangement against a baseline, and review where it helps or creates difficulties.

Should junior employees and new hires always work in the office?

They need timely access to people who can teach them, useful examples, and real feedback. Research on proximity suggests there can be learning benefits to working near teammates, but it does not establish one attendance rule for every newcomer. Coordinate the relevant people when shared time helps, protect mentoring time, and check whether remote or in-person support is actually helping someone become more capable.

How should a company compare the costs of remote and hybrid work?

Keep actual cash spending, the value assigned to coordination time, and employees' commuting time separate. Use realistic workspace and support costs, including whether a lease can change. An hourly value can help discuss time demands, but it is not automatically an extra invoice or a productivity loss. Review customer service, quality, learning, and employee experience alongside the financial comparison.