Not Remotely Correct

Most tech companies run remote offices terribly. There's a better way — but it takes a willingness to question fundamental intuitions.

Quick! Name one tech company’s remote site that’s widely known to be successful, where people wouldn’t immediately move to HQ if given the chance (and HQ’s city wasn’t going Beyond Thunderdome).

If asked, I’d posit this: Google Zurich is broadly known to be a great office to work in. At least as of 10 years ago when I last canvassed this question.

What we learned during COVID was how to unmute when speaking and to claim astoundingly high productivity while coding in pajamas. What we positively did not learn, with all the breathless excitement and endless articles about the shifting future of workspaces, is how to actually run successful remote offices. We made the lives of folks at headquarters easier, but all our newfound knowledge about custom background wallpapers did not improve the experience of work for employees in globally distributed offices.

I started and led Meta’s engineering office in London, which we built from twelve engineers to 600 in five years. I’ve spoken with site directors from Google, Microsoft, Uber, Stripe, and many other companies through the years. Here’s what I’ve learned.

How to Run the Worst Remote Office

If you reflect on many companies you’ve observed, you’ll see examples of all these approaches. To run the very worst remote office:

• Treat it like HVAC for employees, with no strategy around location and purpose

• Scatter a lot of little projects in it, and let people work as one-offs

• Staff it with junior people

• Base all managers in HQ

• Keep product managers and other key influencers in HQ

• Pay everyone less

• Bootstrap it via an acquisition, or center it around one celebrity

• Don’t send culture carriers from HQ

• Hire a very senior, poisonous culture influencer

• Pull the most successful projects back to HQ

• Center the office around geo-specific work (e.g. European features for European offices)

Succeeding in Remote

Most companies do not have the luxury of having remote baked into their DNA the way Mozilla or other born-remote companies do. Instead, most companies grow into needing/wanting remote offices — often reluctantly, because of the obvious overheads.

To do well with remote offices, you need to have:

1. A purpose and strategy. This includes systems-level views of what you want to build (people systems, business systems, and eng systems). And a clear sense of why you’re doing it: why at all, why right now.

2. Broad buy-in. Execs set the tone for how strategic (or not) a remote office is. But managers and senior ICs also significantly affect its likelihood of success.

Classic Approaches

Here are common reasons why companies open remote offices, and common consequences of these approaches.

Acquisition

This is one of the most common reasons a company starts having remote offices: because the acquired people refuse to move. This is “remote by accident” — but it doesn’t have to be bad (even though most cases are).

Pros

• Fully staffed from day one

• In-place leadership structure

• Existing standalone product decoupled from everything else

Cons

• Culture misalignment. Especially difficult with entrenched leadership.

• Us-vs-them thinking. Similar causes.

• Miscalibrated hiring standards.

• Insignificant product (in context of the broader company). You rarely acquire something that is truly central to your company, because if it was, you would have already been building it. You mostly buy additive things meant to augment your business.

There are too many failure cases to list. Successful acquisitions often involve one or more of the following dynamics:

• The acquired business becomes central to the main business.

• Significant acquired staff (especially senior leaders) move to HQ. This has many positives, including the onboarding/osmosis/transference of culture back to the remote office as well as increased representation/mindshare in HQ. It has a subtle downside of signaling that truly important work can only be done at HQ.

• The remote office is given a larger project because of the already-established team.

Retention of Key Employee

This happens more than companies like to admit: a key employee quits to move to another city/country, so execs pitch them on opening a remote office there.

Pros

• The founding member knows the company culture and is trusted by leadership

Cons

• Single-tenant risk. It’s often a disaster when the key employee leaves because the only Pro listed above disappears.

• No development strategy for the office. Shoot → ready → aim.

• Retention of the key employee is executed on a short timeline.

• Odds are good that the key employee didn’t choose their destination city based on HQ’s global development strategy.

Financial Reasons

Tax incentives, wage arbitrage, or geo-concentration of particular skills. Microsoft Dublin is an example of this:

• Low corporate taxes was the impetus.

• Wages were lower, tempting more hiring managers to build teams there.

• EU and other immigration benefits drew localizers for all of Microsoft’s 99 world languages.

• Eventually Dublin was full of translators, such that other companies also started doing translation there.

Companies usually will choose to do this regardless of what I write here, so I won’t say much more. It’s hard to name companies that have built continually-thriving remote engineering offices born of financial incentives.

Immigration Waystation

Visa arbitrage. An example is Vancouver for US H-1B’s, attempted by Microsoft, Meta, and many others.

Pros

• In-company retention is unparalleled. These people cannot, and will not, leave the company.

• Can have good junior talent density. For many of the best students in the world, this is their Ellis Island.

Cons

• Perpetually junior staff. Most people stay only one year until their L-1 or H-1B. And nobody joined because of the chosen city.

• Very high overhead. Onboarding costs dominate because the annual churn is so high. Projects often leave with the people because they leave in droves.

It’s hard to build serious software in an Immigration Waystation. Most companies tend to close them down after several years. Some even have spun up second-attempts (e.g. Microsoft Vancouver).

Geographic Expertise

This is the “For Us / By Us” approach. MSN France was built in France; Microsoft Word (Japanese) was built in Japan.

Pros

• Local expertise

• Local retention. Many of these employees never wanted to leave their country anyway.

Cons

• Hard ceiling to growth & impact, by design. MSN France will only ever be so important to Microsoft.

• Isolation from the rest of the company, also by design. Its reason for existence also becomes its reason for isolation.

The very thing that seemed a strength in your first year quickly becomes the reason everyone leaves in the third. It’s hard to name software companies that have had sustained success with this, except in cases where the product is so radically different as a standalone powerhouse in its host country.

Field of Dreams / Altar Call

If you provide HVAC and free food, they will come.

One of the most common reasons given for opening a remote office: hire talented people. This approach often tilts heavily towards operational decisions (e.g. what city? which office building? what compensation?) and away from strategic planning (e.g. why now? how many offices is too many? which projects, which leaders?)

They come at first, usually in droves, because you tap into a location’s long-pent-up demand to join you. But this initial success is deceptive. Two to three years after recruiters have collected their bonuses, retention issues begin to mushroom because everyone senses a lack of strategy for the office and a lack of career growth opportunities for themselves.

Building Great Remote Offices

Any of the above Classic Approaches can be made to work. They aren’t doomed to failure if you know the pitfalls. Here are my recommendations to bias towards success.

Who

The single-biggest influence on the initial success of Meta London came from conversations I had with David Singleton — then a Google London Site Lead and eventually Stripe’s CTO — where we discussed the reasons for various Google remote offices either thriving or withering.

It’s ultimately all about whom you send. They should:

• Already be influential and respected in HQ. The mindshare and impact of this social capital is invaluable, both in landing significant projects and in retaining trust across offices.

• Commit to staying two years in the new office. I felt this was crazy when I first heard it about Google Beijing, but it significantly bolstered Meta London’s success to insist on a landing team that would stay two years.

• Be at least six, ideally twelve people. A super-pragmatic reason is that you want two full interview loops; but really, this is about the titration rate of new hires and the resultant dilution of culture you risk when you send too few.

I brought twelve people with me to launch Meta London’s engineering office, after interviewing nearly 40 volunteers. I polled their peers and managers regarding how well they represented the company culture, their track record of success, and how much others trusted them. I asked them each to bring a project with them so that we could seed the office with work from day one. The composition, size, commitment, and social capital of the initial landing team was the single most important determinant of Meta London’s success.

What

Vision: Build a small number of large projects of company-wide import, with tight local cohesion, run and managed cross-functionally on site.

Portfolio, Not Project

You shouldn’t run a remote office like a startup. You should run it like a VC. When it comes to what to build in a remote office:

• Treat the entire office as a portfolio of projects. Optimize for portfolio growth, not for individual project growth.

• The worst team is a zombie team. You either want teams to be growing fast or you want to proactively shut them down and reallocate their people to the highest growth teams. A team that grows 20% a year is a remote office’s death — not failed enough to shut down, not successful enough to anchor the office.

Loose Coupling, Tight Cohesion

Good software architecture principles are exceptionally helpful in this case. Build teams that are:

• Loosely-coupled, both communication- and management-wise. The greatest expense of a remote office is its engineering efficiency overhead. Strive for well-defined, slow-to-evolve boundaries of responsibilities across offices.

• Tightly cohesive. It’s best to have teams operating cross-functionally in the same location, managed by local managers.

Global, Not Local

If you give a remote office something specific to their country or geography, you’re pre-capping their maximum impact at the relative importance of that geography to your business. Similarly, if you put non-essential projects in remote offices, then the offices themselves are, by definition, non-essential. Both hiring and retention will be issues.

When

If you continuously attract as much of the world’s top talent to your HQ as you need, you should stay geographically consolidated. Apple was successful with this strategy in Cupertino because they grew over decades. If, however, you are unable to attract as much talent as you need in one central location, you need to either slow your hiring, lower your bar, or open a remote office in a talent-dense geography.

There’s no magic to When, but here are some ways to think about it:

• Culture carriers, project cardinality, and leadership are all rate-limiting. You need enough trusted people to be available and willing to move, under strong leadership. You also need enough significant projects that can be factored out and executed remotely. Having too few projects limits hiring, poses concentration risk, and builds unnatural momentum to keep projects alive that should be canceled.

• Remote offices are not one-way doors. Companies treat these decisions as irrevocable, but it’s beneficial to be antifragile about them. You of course don’t want to randomly open offices all over the place. But you should treat them with the same gravity, planning, and willingness to be proven wrong as you do when embarking on any significant company project.

Where

Here are tradeoffs I’ve observed about various cities:

• Attractiveness. It’s easier to draw people to New York or London than to DC.

• Language. People are more willing to move to London than to Berlin because they either already speak English, or they value it as a portable skill to learn. This is one of the things that make Singapore interesting if you want to be in Asia.

• Operational cost. It might be helpful to separate traditional opex from salaries. For instance, London is very expensive in real-estate and other costs, but is cheaper in salaries than the US.

• Ease of immigration. Vancouver and London are both easier than the US, for example.

• Availability of senior leadership. This usually lags local tech offices by a few years. When I worked in Shanghai in 2008, it was much easier to hire junior people than in the US because of supply/demand dynamics, but much harder to hire E7+ people because the local industry hadn’t matured enough. London was similar in 2012. Seattle is unusually strong on this axis.

• Time zone. Pacific Time people often think of New York and London both as “far” places, but New York has 5 hours to overlap with SF every day, whereas London has only one. This difference is massive, compounded daily. India is terribly positioned: there is no time day or night where you can have a meeting that doesn’t significantly impact one or both sides.

Why

Next to who you select for the landing team and to lead the office, the purpose of the office has the largest influence over its long-term success. In my experience at both Microsoft and Meta, senior leaders were often not on the same page about why an office even exists. Divergence of buy-in begets divergence in strategy, resulting in an office that lacks coherence. Employees can feel this, and it reflects in both morale and retention.

Remote offices thrive when senior leaders throughout the company make a unified bet to make the office succeed. To make an office successful, attention, projects, and people need to be allocated strategically and supported at the highest levels in the company.

How

This is the Open Mic Night section where random thoughts large and small go in no particular order.

When I stir everything in a pot, my rank order of offices for a Silicon Valley company to open would be: Seattle > London > Singapore. This strategy favors worldwide coverage in attractive, English-speaking locations where talent is already plentiful or can be convinced to move via flexible immigration, where laws are predictable/favorable, in decreasing order of time-friendliness. Bangalore/Hyderabad and Shanghai/Beijing fall significantly lower. Other US cities increase sharding without significant incremental benefit over SF/Seattle.

You need one or two people on the landing team who are great company-culture evangelists. And you need one or two really fun people. With them in place, others can easily support the goodness going on; without them, you run an office that can feel staid.

Metrics to measure success, beyond obvious traditional measures:

• Top metric: NPS according to HQ leaders and ICs. If you ask anyone in HQ, they should say, “<remote office> is hiring great people and doing impactful work.”

• Cross-site flow. People vote with their feet. A great office should attract as many HQ people to it as it sends to HQ.

• % ICs locally managed, and eventually % local managers who manage HQ ICs. The latter is a measure of impact/influence equivalency.

• Retention rate, both in-company and in-site.

• Career velocity and asymptote. Median rate of progress through both junior and senior levels. What’s the highest level anyone’s been promoted into?

• Level pyramid. Is it distorted relative to HQ? Is that distortion intentional?

The earlier the hire, the more disproportionate their impact on site culture and success. There’s a titration rate that you need to balance carefully in order to promulgate the culture you want. A few visible/vocal people set the entire tone.

Finding a replacement site lead can be difficult, because you usually need a replacement in 2–3 years, whereas local hires are often not ready to step into those roles unless you’re in a geography dense with senior leaders. It’s better, for this role, to hire from within.

Whom the site lead reports to is an important signal to candidates. I found it much easier to recruit candidates, especially E8+ folks, during the years I reported directly to Meta’s CTO. It’s also a strong signal within the company regarding the relative importance of the site. Lastly, it’s difficult to have a broad-impact site if its lead reports into a specific product/domain; the site becomes a vassal of a team, not a pillar of the company.

A single wrong move in hiring a senior local leader can be disastrous. Meta Seattle hired such a person, and he poisoned the culture in an insidious way. There are several reasons senior hires are especially tricky in a nascent site:

• People naturally want to follow and please them, so they’re disproportionately impactful

• They usually manage large teams, so extracting them is disruptive

• Triangulating their negative influence is difficult because they can dominate the interface to HQ leadership

Meta tried a few sites with “co-leads.” This has analogously serious downsides as a company having co-CEOs. People need one “spiritual leader” who sets the tone of the site. Both internally and externally, there needs to be clear accountability and agency to make binding decisions.

Each site, if healthy, will develop its own traditions and variations on HQ culture. It’s like visiting Iowa: you know you’re definitely in America, but you can also tell it’s Iowa.

One of the most damaging things HQ can do to a remote office is move a project back to HQ when it gets big or important. There are tempting reasons to do so: more available people to staff hypergrowth and closer interactions with execs commensurate with its importance. But it both reinforces an attitude that truly important things can only be done in HQ and also suggests that the fruits of success are to lose the thing you birthed and nurtured.

Like a mall, sites benefit from anchor tenants. Meta Workplace was the first project in London that was:

• invented there

• cross-functionally staffed all the way to the director level

• of obvious business impact to the whole company

Having an anchor tenant is huge for recruiting and morale.

Recruiting is not the most important thing for a remote office.

Recruiting is not the most important thing.

It’s not. Everybody treats it as if it is. But they’re wrong.

You seed an office with great people, a strong leader, and a few fledgling projects. You bet on and support its teams. Great work attracts great people. The site, built well, sells itself. Lead with the strategy and purpose. Recruiting will naturally flow from that. Don’t invert the causality.