10up was designed from the ground up to be a distributed agency. At the time, I had a really good reason and a rather lame reason.
The good reason: I’d spent the last 5 years in leadership positions at consultancies, a “hub and spoke” model with 3 offices around the country and a few one man “stations”, and one company with a centralized office adding a space in a larger city to expand its talent pool. At both shops, I played a central role in recruiting production talent. While building the latter team, I was beginning my immersion into the WordPress community, notably the WordCamp phenomenon, where I met amazing talent all over the country. I became convinced that the idea that one could build the best possible team while limiting ones self to one – or three – 50 mile geographic living radii was faulty.
Costs & Benefits
This is not to say I was oblivious to the benefits of in person collaboration, having worked with both local and very remote clients, some of whom were involved in the technical builds, as well. It was to say that the costs of remote collaboration had fallen dramatically over the last decade: from reasonably good and incredibly cheap home video conferencing and collaboration tools, to the rising dominance of ever-connected smart phones, to access-anywhere based collaboration tools like Google Docs and Basecamp, the barriers to good remote collaboration were falling quickly. There was also precedent to model, with 37signals (now Basecamp) and our community’s own Automattic leading the remote charge.
At the same time, the benefits of “hiring anywhere” were rising rapidly in our space. I saw an increasingly sub-specialized web technology industry where that sub-specialized talent was becoming even more diffuse and in-house training was becoming more difficult. Put in simpler terms: the number of web technologists, broadly, was rising steadily, but the number of areas requiring specialization was rising much faster. The net effect: it was harder to find someone really good at a specific, high skill task in one geographic area, and the increasing complexities of our work was making training and mentorship increasingly more costly, especially when you consider that the average job-span in the technology sector, and other youth dominated industries, is about 2-3 years (and hence the return on training investment much smaller).
I also mentioned that I had a lame reason: and while it feels unbecoming to say that in a time of great personal change, I didn’t know where I wanted to live in 6 months or 3 years, it’s actually a pretty good reason when you consider how commonplace that is in the labor market. From military spouses to talent that hasn’t settled down in their personal life, the distributed model is incredibly attractive to a significant fraction of the talent market that is increasingly transitive, be it by necessity (traveling families) or lifestyle preference.
There are other benefits to the distributed model:
- Commuting robs the average worker of 38 hours a year – almost a full work week (and lord help you if you’re in Los Angeles).
- Forceful work interruption by want of “swinging into a cubicle to chat” probably robs as much productivity as happenstance moments of in-person magic create.
- Knowing that my best working hours in my twenties were often between 9 and 10:30 pm, and my worst hours between 3:30 and 4:30, I know first hand that arbitrary “everyone here between these hours” policies are often counterproductive. Put differently, it puts the focus on hours worked, instead of hours in an office.
- It may come as a shock that there are many talented and introverted engineers who see a quiet, private work place as a huge benefit.
- While a virtual water cooler chatroom is admittedly less organic than a literal water cooler (though increasingly less so in the SMS-era, or whatever we call mobile messaging), it’s also much better. Less prone toward negative gossip when written in a public forum, and more importantly: chat history. Did you miss the company chatter while you were focused in on your deadline? Scroll up.
- Brand ambassadors and representatives all over the country, and world.
And yes, there are very clear benefits to the in person model. Emotive and subtle communication, and hence empathy, occur and are understood much more naturally. It is still easier to brainstorm and whiteboard ideas. The buzz of the workplace (as compared to the siren song of the Playstation in your living room) is somewhere between helpful and essential for some – effectively the inverse of the benefits of remote to the introverted.
Its for these reasons that 10up gets everyone together once a year in the same spot, focused on collaboration, offset by about 6 months from team meet ups, with semi-frequent encounters with varying team mates at events, conferences, and in person client meetings throughout the year. In fact, as a distributed shop, we end up deliberately scheduling extended, discreet time for in person collaboration and team building in ways that many physical shops never bother to prioritize beyond a monthly lunch and learn and picnic.
The “blended” model
I’ve seen this over and over again: a shop grows to about 20, runs out of local talent, doesn’t want to open an office somewhere else, can’t convince (or afford to pay for) talent to move, and so it looks beyond its borders. Maybe a key employee needs to move for family reasons, and the company wants to find a way to retain them; let’s not pretend this is any different. The company has decided they can’t find a comparable talent for comparable cost – which includes experience with the company itself – locally.
The result, simply put, is a central office where the company started and most key personnel work, with some number of distributed employees. Almost always because at some point the primary benefit of a distributed workforce – talent access – forces it.
As with most things thrown into a blender, the result is often messy. While you might be inclined to call this “blended” model the best of both worlds, I see this as a “compromise” model, often the worst of both worlds. That’s not to say that the worst option can’t and doesn’t work; indeed it can and does: it just means I think it’s the least effective model.
The lack of full distribution creates (literal) water cooler and happenstance conversations in the central office, where it’s literally impossible for remote employees to participate in the conversation, instead of pushing conversation into a shared “online” forum. Along those lines, emotive relationships are built within the central office, and infrequently developed in equal proportion with the broader distributed team. Team meetings where everyone joins on a webcam are replaced with awkward “being the guy on the projector screen staring down at the half of the office team that you can actually see.”
The result is invariably a first and second class of employees, with the local crew around the conference room table, and “those other folks you have to dial in and lean forward to hear on the speakerphone.” Hiring key executives that are also fully distributed can mitigate the imbalance, but happens infrequently. In part because those would-be executives would rather not be the one executive not in the headquarters.
Hub & Spoke
The hub and spoke model implies a central office, usually – but not necessarily – home to “corporate” or the final decision maker(s) – the hub, and several typically smaller offices or even one person satellites (akin to the blended model) – reached with spokes. WPTavern wrote about this today, interviewing Chris Lema in the wake of his joining a hub and spoke agency, which inspired this post. It occurs to me now that this might just be part of Chris’s strategy to get me to blog.
The hub and spoke model often exhibits many of the same attributes as the “blended” model, at scale. Strong cultures and relationships tend to form in each office, often creating micro-cultures. That’s not necessarily a bad thing; especially if executives are evenly dispersed across offices (putting all of the executives or shareholders in one of the offices risks an ugly “workers vs managers” culture I’ve personally witnessed). Pockets of culture and bonding is light years better than singular isolation from a centralized culture, not to mention inevitable with growth, anyhow. And you do get the happenstance benefits of in person collaboration, even if its in silos that fall short of being under a single roof.
And yet, I respectfully find this conclusion (in the WPTavern piece) incredibly faulty, in the context of comparing workforce models: “I think we’ve tried to thread the needle on this to find the situation that gives us the least amount of drawbacks with the most amount of potential.”
To the extent I define the primary benefit of a distributed workforce as finding the best talent wherever they may be, what I hear in that is “we’re better than a central office because we’re limited to 6 geographic radiuses for finding talent, instead of 1”. Well, sure, better than one office, I guess, but that’s a bit like saying your high end SUV is the best of both worlds between an economy car and an armored tank. If the goal is not to be blown up, it’s not really.
If the goal is finding the best talent wherever they may be, by my calculation, 6 officers covers – generously – about 400 square miles of talent, instead of 57,308,000 square miles of talent around the world. And yes: while it’s not fair to count Wasilla Alaska’s square mileage as equal to New York City’s, the point very much remains. 10up has found amazing talent in locations as remote as a Canadian island, a small town in Idaho hours from an airport, and a remote part of Vermont. One of our best engineers, who left a senior engineering position at Amazon to work for 10up, is traveling the globe! Oh, and yes, a Director in Anchorage, Alaska.
What makes the “best of both worlds” claim even more surprising is the centralization of different talent in each locale. You actually might convince me that Vegas is uniquely suited to video and promotion, but I don’t accept that New York City is uniquely suited to Ruby engineers, that Phoenix has a concentration of the best designers, or that something in Denver’s water makes it uniquely suited to “advanced development” (although we do have a few great engineers out there, come to think of it). How does this touch the talent benefits of full distribution?
Its argued that putting like talent in one spot enables these teams to go “faster”. I don’t dispute that this is an ideal. I simply and respectfully argue that finding better talent by not limiting geography yields the most thrust, and often, a diverse, more global perspective also offers tangible creative benefits. Or, what one might see as getting some of the benefits of each model, might also be seen as getting the full benefit of neither model.