Hybrid Work Lessons from Atlassian, Dropbox, Remote.com

Atlassian, Dropbox and Remote.com all went distributed in 2020, and none of them has quietly reversed it since. Read their published numbers side by side and the same lesson shows up three times: letting people choose where they sit is the easy part. Rebuilding how decisions get made, written down and reviewed is the work.

That gap is why so many hybrid policies read well and land badly. A policy that only answers "how many days in the office" has not answered anything a distributed team actually struggles with.

TL;DR

• Atlassian's Team Anywhere lets staff pick their location daily, and the company now spans 11,000+ people in 13 countries and 14 time zones.
• Dropbox's Virtual First is the strongest hiring evidence of the three: applications per role are close to seven times pre-2020 levels, and 60% of staff live outside big tech hubs.
• Gallup's data shows hybrid settling rather than collapsing, at roughly 2.3 office days a week.
• The quiet finding: on-site workers whose teammates sit elsewhere doubled from 13% to 27%, so being in an office no longer means being together.
• Write the policy around documentation, decision rights and gathering rhythm, not around a day count.
• Hiring is where most of this breaks. Async judgment and clear writing do not show up on a resume.

What is Atlassian's remote work policy?

Atlassian calls it Team Anywhere. Employees choose where they work each day, including full time from home, and the company hires anywhere it holds a legal entity. Atlassian reports 11,000+ employees across 13 countries and 14 time zones, with 40% living more than two hours from any office.

The part people miss is that Team Anywhere is not an office-attendance rule at all. There is no required number of days. Instead the company sets a gathering rhythm: teams come together in person three to four times a year, and its 11 offices are positioned as places to gather rather than places to sit daily.

Atlassian Team Anywhere remote work policy: what changed between 2025 and 2026

Compare Atlassian's own account of the policy over time and one thing has visibly moved: the gathering cadence. In its January 2024 review of the program's first 1,000 days, Atlassian described intentional team gatherings roughly three times a year. Its current description sets that at three to four times a year, with offices reframed around those gatherings.

That is a small edit with a big idea behind it. The 2024 review found that a gathering lifted people's sense of connection by 27%, and that the lift lasted four to five months. Four to five months of half-life, and a calendar year, is exactly where three gatherings a year stops being enough and four starts to make sense. The policy did not change because sentiment shifted. It changed because the decay curve said so.

The same 2024 review reported that 92% of Atlassian staff said the policy let them do their best work, and 91% called it an important reason they stay. Those figures are two years old now and Atlassian has not restated them, so treat them as evidence the model held early, not as this year's score.

How does hybrid work look across the three companies?

All three are remote-first rather than hybrid in the usual sense, but they differ on where the source of truth lives. Atlassian optimises for choice plus scheduled gathering. Dropbox splits the week between deep work and structured collaboration. Remote.com pushes almost everything into writing and treats meetings as the exception.

The knowledge layer differs at each company: Jira, Dropbox Dash and Notion each index a different slice of the work.

Notice what none of them did. Not one built the policy around a fixed office quota. They each picked a coordination mechanism first, then let location follow it. A team that has to be in a room to make a decision is not a distributed team with a location policy, it is a co-located team with a commute exemption.

What does Dropbox's Virtual First model prove?

Dropbox is the clearest case that going remote changes who you can hire. In its January 2026 review of Virtual First, Dropbox reported that applications per role are close to seven times higher than before the model, that 60% of employees now live outside large tech markets, and that attrition is the lowest in company history.

The talent-pool effect is the headline, but the meeting data is the more useful lesson. A Dropbox pilot that deliberately cut meetings produced a 27% drop in weekly meeting hours, about 1.8 hours back per person, 20% fewer meetings, and a 28% rise in coding time. Eighty percent of participants judged the pilot effective.

So the win did not come from the policy. It came from a team running an experiment against the policy and measuring it. That is the repeatable bit, and it is the bit most companies skip.

Why do most hybrid policies fail?

Because they answer the wrong question. Most policies specify days in the office. Almost none specify how a decision gets recorded, who is allowed to make it without a meeting, or what happens when half a team is asleep.

The market data backs this up. Gallup's Q2 2025 survey found hybrid working slipped from 55% to 51% of remote-capable US employees, with hybrid staff spending 46% of the week on site, about 2.3 days. Employees saying their schedule is entirely their own call fell from 37% to 34%. That is a drift toward employer control, not a retreat from hybrid.

Here is the number that should reshape the policy. Among fully on-site remote-capable employees, the share who say their team is spread across different locations climbed from 13% in 2023 to 27% in 2025. Bringing people back to a building does not put them back with their colleagues. Half the "return to office" argument is answering a problem that the office no longer solves.

Pro Tip: Before you set a day count, audit one real decision from last quarter. Where was it made, who was in the room, and where would a new joiner in another time zone find the reasoning today? If the answer is "in a call, and nowhere", a day count will not fix it. Documentation rights and decision rights come first. Consider how quickly communication channels multiply as a team grows, and the case for writing things down gets stronger.

The honest caveat: async-heavy models have a real cost. Written decisions are slower to reach when something is genuinely ambiguous, and new joiners take longer to find their footing without ambient context. Remote.com's answer is heavy documentation plus a buddy for new starters. Dropbox's answer is offsites. Neither pretends the cost is zero, and a policy that claims no tradeoffs has not been tested.

How do you hire for a distributed team?

Test the behaviour the policy depends on. If the model runs on written decisions, then writing clearly under ambiguity is a job-critical skill, not a nice-to-have, and it is close to invisible on a resume and easy to fake in a friendly video call.

The skill mix is shifting anyway. The World Economic Forum's Future of Jobs Report 2025 found employers expect 39% of workers' core skills to change by 2030, with analytical thinking, resilience, flexibility and agility at the top of the list. Those are exactly the traits a distributed team leans on hardest, and exactly the ones a CV bullet cannot evidence.

This is where the Testlify Multi-Signal Talent Evaluation Model applies. It combines several role-relevant signals, including skills assessments, work simulations, interviews, reference checks and reviewer feedback, so a shortlist rests on more than one polished conversation. One signal is fragile. Several pointing the same way is a decision you can defend.

For a distributed role, that usually means a written work sample scored against a rubric, a communication assessment, and a structured interview where more than one reviewer scores independently. Teams building this out often pair it with workforce planning for distributed teams and with recruiting across time zones and cultures, since the policy, the plan and the hiring bar have to agree. Compliance is its own thread: rules that follow the employee, not the office, decide where you can legally hire at all.

Ready to test for the skills a distributed team actually runs on? Book a demo and see how a multi-signal shortlist comes together before the first interview.

Key Takeaways

• A day count is not a policy. None of the three companies built their model around office quotas, because the quota answers attendance and the problem is coordination. Write down how decisions get made and recorded before you write down how often people show up, or the policy will be relitigated every quarter.
• Gathering rhythm should follow measured decay, not tradition. Atlassian's connection lift from a gathering lasted four to five months, which is precisely why three or four gatherings a year works and an annual offsite does not. Measure your own half-life instead of copying a cadence.
• Remote widens the funnel more than it changes the work. Dropbox's applications per role sit near seven times pre-2020 levels with 60% of staff outside major tech hubs. If your hiring bar is not built for that volume, the policy creates a screening problem before it creates a talent advantage.
• The office no longer guarantees togetherness. The share of on-site employees with geographically split teams doubled from 13% to 27% in two years. Mandating attendance for collaboration that happens across time zones anyway buys commuting, not collaboration.
• Experiments beat policy rewrites. Dropbox's meeting pilot returned about 1.8 hours a week per person and lifted coding time 28%. Small measured changes moved more than a policy revision would have, and they are far easier to reverse when wrong.
• Hire for the mechanism you chose. An async model depends on written judgment, so assess it directly with work samples and structured, multi-reviewer scoring rather than inferring it from a resume or a warm interview.

FAQs