Fewer office days, more reasons to stay

Imagine judging a workplace policy by how many desks are occupied. You might miss a different question: who still wants to work there? In a randomized trial at Trip.com, offering some home working reduced departures. The offer mattered: employees could use two home days each week, but typically used about one. The study tested access to flexibility, not compulsory absence.

What supports the answer?

  1. Reported finding: Fewer employees left
    Reported finding

    Among 1,612 employees, departures during the 2021–2022 trial were 4.8% with the hybrid option and 7.2% without it: 2.4 percentage points lower, or one-third in relative terms.

    Scope and conditions: The offer allowed home working on Wednesday and Friday; assignment used odd or even birthdays.

    Source passage · 2024

    See the randomized offer, absolute quit rates and actual take-up. The result concerns eligibility for hybrid working.

    Effects on employee retention, Fig. 2; Methods: Randomization and Employee characteristics and balancing tests. About the source and its version
  2. Reported finding: Performance grades held up
    Reported finding

    The first review period showed similar performance grades. Equivalence tests ruled out average differences as large as half a grade in either direction on the five-point scale.

    Scope and conditions: That does not establish identical performance or measure every aspect of work quality.

    Source passage · 2024

    The first review covers July–December 2021. The equivalence bounds were ±0.5 grade; subsequent reviews followed rollout to the former control group.

    Employee performance and promotions, Fig. 3; Methods: Null results; Discussion on rollout after the experiment. About the source and its version
  3. Reported finding: Connections can change too
    Reported finding

    A separate analysis of 61,182 US Microsoft employees estimated that company-wide remote work made collaboration networks more inward-looking, with fewer links across groups.

    Scope and conditions: This was a pandemic-era natural experiment, not a trial of two home days. It did not directly measure productivity or innovation.

    Source passage · 2021

    The analysis uses a modified difference-in-differences model. Its causal interpretation depends on assumptions including parallel trends. The October 2021 correction fixes a direction label in Fig. 1.

    Pages 45–47: collaboration networks and Fig. 3; pages 50–52: discussion, limitations and Methods. About the source and its version

Where this explanation stops.

  • Trip.com studied graduate employees in two divisions in Shanghai; interns and probationary newcomers were excluded. Later follow-up came after the hybrid policy expanded to everyone.
  • Both studies involved company insiders as authors. Trip.com co-author James Liang disclosed an equity interest; most Microsoft authors disclosed financial interests in Microsoft.

What to take with you

Atlas interpretation: ask what a policy changes, for whom, and over what period. Keeping employees, assessing their work and connecting teams are separate outcomes. Evidence about one should not silently stand in for all three.

Follow the answer back to its sources.

Peer-reviewed randomized workplace trial · Source date: 12 June 2024

Hybrid working from home improves retention without damaging performance

Nicholas Bloom, Ruobing Han and James Liang

Nature 630, 920–925 · Version of record, 12 June 2024

Online publication: 12 June 2024; journal issue: 27 June 2024. Research dates differ from this Atlas entry.

DOI: 10.1038/s41586-024-07500-2

Source passages, license and conditions ↗

Peer-reviewed observational natural experiment · Source date: 9 September 2021

The effects of remote work on collaboration among information workers

Longqi Yang, David Holtz, Sonia Jaffe, Siddharth Suri, Shilpi Sinha, Jeffrey Weston, Connor Joyce, Neha Shah, Kevin Sherman, Brent Hecht and Jaime Teevan

Nature Human Behaviour 6, 43–54 · Corrected publication 2021; journal issue January 2022

Published online 9 September 2021. Author correction, 5 October 2021, fixes the direction label in Fig. 1: https://doi.org/10.1038/s41562-021-01228-z. Journal issue January 2022.

DOI: 10.1038/s41562-021-01196-4

Source passages, license and conditions ↗

How this version was checked

AI-reviewed by a separate editorial agent · 22 October 2026. The approval applies to revision 1, its text, visuals and source references.

Why this piece

Chosen to explore a familiar workplace disagreement through measured outcomes. The evidence concerns particular employers and arrangements; it does not establish a universal office schedule.

AI-reviewed by a separate editorial agent · Ask the editor

What changed, and why.

This history comes from saved content versions. A new revision can be an editorial change; it does not by itself mean that the underlying facts changed.

Latest reviewed revision

First recorded version · revision 1

First Atlas account of hybrid work, retention and the limits of workplace evidence.

Recorded 22 October 2026 · reviewed 22 October 2026 · AI-reviewed by a separate editorial agent

No earlier reviewed assessment is available to compare. No change in the underlying facts is being claimed.

Available version records (1)
  1. Revision 1 · First recorded version · reviewed

    First Atlas account of hybrid work, retention and the limits of workplace evidence.

    Recorded 22 October 2026 · reviewed 22 October 2026 · AI-reviewed by a separate editorial agent

    Answer in this version

    The Microsoft study examined connections between workers, rather than testing a weekly hybrid schedule. It compared changes among previously remote and office-based employees, using assumptions about how the groups would otherwise have changed. That helps address pandemic effects, but does not make the study a randomized trial.

    What remained unresolved

    An informative next experiment would vary shared office days across teams and follow new collaborations, mentoring and work quality over time. That is a research proposal, not a result of these studies.

    Takeaway and scope

    At Trip.com, the option to work from home helped employees stay.

    A hybrid-work trial at Trip.com, with a separate study of fully remote work at Microsoft.

    Fewer employees left

    Among 1,612 employees, departures during the 2021–2022 trial were 4.8% with the hybrid option and 7.2% without it: 2.4 percentage points lower, or one-third in relative terms.

    The offer allowed home working on Wednesday and Friday; assignment used odd or even birthdays.

    Performance grades held up

    The first review period showed similar performance grades. Equivalence tests ruled out average differences as large as half a grade in either direction on the five-point scale.

    That does not establish identical performance or measure every aspect of work quality.

    Connections can change too

    A separate analysis of 61,182 US Microsoft employees estimated that company-wide remote work made collaboration networks more inward-looking, with fewer links across groups.

    This was a pandemic-era natural experiment, not a trial of two home days. It did not directly measure productivity or innovation.

    Sources in this version