How we Communicate at Automattic

When I talk to people about my job at Automattic, most of them can’t really comprehend how it can possibly work since it’s so different to what they see as a ‘normal job’.

Distributed Not Remote

Automattic, as it stands, employs 480 people across 45 countries, in almost if-not-all timezones, and everyone works in a distributed manner, with no central office and no core business hours.

One common theme that people ask me about is about communication. They may either say something like ‘oh, it must be hard to communicate‘, or ‘you must have many late night, or early morning, conference calls you must attend‘, which of which are both typical responses on the discussion of remote working.

But since we’re distributed, I actually don’t have any times where I need to be on a ‘call’, our team has no live video chats, and I feel our level of communication works better than any ‘office’ I’ve previously worked in.

Asynchronous Communication FTW

I believe the reason that communication works well at Automattic is that everyone is distributed, so on an level playing field, and this means they can live in any timezone in the world. This also means any communication needs to be asynchronous.

What do I mean by this? When you have a phone call and each person is talking to each other at the same time, this is synchronous. A video call, or in-real-life (IRL) conversation, is the same.

But when you are emailing, text messaging, iMessaging or even Tweeting with someone, and you take turns sending each other messages, this is asynchronous. Asynchronous communication doesn’t mean it can’t be happening at the same time, but it does allow communication to span over time.  Whereas synchronous communication can only happen in the here and now.

Why is asynchronous so important?

If everyone is distributed across every timezone in the world you can’t possibly engage with everyone synchronously if you don’t expect them to work reasonable and convenient hours. For example, as I start my Saturday morning I have colleagues who are still working on their Friday afternoon.

This is why we don’t have video or conference calls (for the most part1) at Automattic.

Even during my hiring period, all my communication was asynchronous. In my ‘interview’, which was conducted via a text chat, my future team lead would ask me a question, and since I would often be sleeping, or doing something else, I could respond any time I liked (when he may have been sleeping). It was like that throughout my hiring period and trial; so I knew exactly what to expect should I be hired.

Ways to Communicate Asynchronously

Whilst email is an asynchronous communication method it is a terrible one at that. Ironically it is often over-used (abused?) in ‘office’ situations where synchronous communication is possible. One reason email is bad is that you can’t ‘opt-in’ to a work email chain going on, so for this reason it is often the case of additional people being CC’d into the conversation ‘just in case they’re interested or FYI’ (which was probably why GMail invented the awesome yet sad Mute button).

I’ve never met someone who has absolutely loved communicating via email.

No Email at Automattic

At Automattic, we don’t use email to communicate. That’s not to say we don’t have an email account, we do, it’s just that I would very rarely use it to communicate. I could count on my hands how many emails I have sent to someone in almost a year.

So if email is terrible for communication, even though it’s asynchronous, what’s a better alternative?

Three Levels of Communication

One of the problems about email communication is that people try to use it for different types of communication, from a quick email to your colleague about lunch, to discussing a new HR policy, or being informed about new starters/leavers.

I’ve held in my mind an idea about different types of communication for a while now, and when I saw how communication works at Automattic it perfectly fit into this model:

Communication at A8C

Like most things, it’s easier to explain this model with some examples.

An example of the three levels of communication at Automattic

Say I have a new idea about something at work, for example, I think we should automatically check for JavaScript console errors during our e2e automated test execution. I might start with an asynchronous conversation in Slack2 about this, just mentioning it and seeing if anyone has any ideas. Someone might mention they saw a blog article about that recently, and post a link to it. I’m immediately ahead before I started that conversation since I now have a head-start on how to achieve this.

I go about my ways of working on this and having resolved a few different issues along the way through conversation, I am now ready for discussion on my idea. At Automattic we make extensive use of internal sites called P2s3 which are a way to quickly post an idea internally for people to read and have threaded discussions. So for example I could post all the details I have about my idea so far, and it’s via this I learn about another approach that’s currently taking place by a different team using a service called Sentry.

Taking all the discussion and feedback into account, I may choose to add information on JavaScript console logging to a new or existing article on our knowledge base called The Field Guide4. This is the guide to all things Automattic and contains only the publication of information, not discussion. It’s still kept very up to date by allowing everyone edit access to any part of it (much like a wiki) – and each page shows the people who have edited it the most.

Another example of the three levels of communication in my personal life

I am (currently) fascinated by ‘continuous dogfooding‘ and whether this can dramatically reduce the amount of typical pre-release testing we conduct on a software product. So I have a discussion one day over lunch with a friend (who has just moved back to Brisbane – yay!) and we discuss dogfooding and I hear for the firs time about an example of ‘forcefooding‘ where dogfooding goes wrong and employees resent having to use unstable software. The darkside of dogfooding.

I think about that for a while and decide to write a new blog post on the topic to see if anyone else has similar tales.

I get lots of feedback and stories about dogfooding and forcefooding and decide to write and release a small eBook on the topic.

All communication levels are asynchronous and opt-in

The important thing to realize about this communications model is that any level of communication is opt-in.

I can opt-in to joining whichever Slack channels I choose, the same applies for ‘following’ a P2 site’s articles5, or choosing to read the Field Guide. Yes, there are ‘essential’ P2s and Slack channels, but the choice is ultimately up to the individual what they join/follow/read, unlike included on email threads by various people.

Communication is Oxygen

Our company creed includes a line about communication (emphasis added):

I will never stop learning. I won’t just work on things that are assigned to me. I know there’s no such thing as a status quo. I will build our business sustainably through passionate and loyal customers. I will never pass up an opportunity to help out a colleague, and I’ll remember the days before I knew everything. I am more motivated by impact than money, and I know that Open Source is one of the most powerful ideas of our generation. I will communicate as much as possible, because it’s the oxygen of a distributed company. I am in a marathon, not a sprint, and no matter how far away the goal is, the only way to get there is by putting one foot in front of another every day. Given time, there is no problem that’s insurmountable.

This encourages us all to communicate as much as possible, which is definitely essential. But I would add to this, that we should communicate is the best possible way, taking into the account the different forms of communication we have available to us.

What communications challenges does this raise?

Whilst I believe the communications model I outlined is a good way to communicate, this doesn’t mean that it’s not abused.

I no longer use Twitter but I remember when I did that people would have almost endless (and circular) discussions on Twitter that would have best been condensed down into a blog post with discussion happening from there. This is because I believe writing a blog post takes a lot greater mental effort than a tweet, and the act of writing a blog post, without continuous feedback/distractions, leads to clarity of thought about a topic.

This may be one reason why I don’t see as many blog posts anymore, because much of these discussions are happening on a much more transient platform like Twitter.

It’s sad because blog posts are a more packaged form of information that’s easily accessible to a wider audience (web users/readers) than just people who follow you on Twitter.

I have seen the same thing happen in a work context where Slack can be overused as a discussion tool, which it’s not, as it’s hard to rely upon its history since it can be so noisy with conversations happening, and there’s (as I write this) no concept of discussion threading. There’s a saying we have “P2 or it didn’t happen” which means that any discussion that takes place in Slack or otherwise should be documented in a (quick) P2 post, so that it’s easy accessible and visible outside the context of that particular Slack conversation.


By thinking about how your teams communicate and using the right tools in the right ways means that organizational communication can be effective, even for a company with employees distributed across the planet.

And next time you’re deep in discussion about something with someone on Twitter or in real life, have a think about how you can take this to the next level by condensing your thoughts into a blog post which you can share with the world for meaningful discussion.

[1] I have seen that some teams at Automattic do have a semi-regular video hangout – but this is atypical. We also have a monthly town-hall with our CEO which is video-streamed (with questions via live text chat), but this is mostly during either American or European business hours, so I typically watch this asynchronously after it has happened.
[2] Slack is a ever-popular instant messaging platform for teams/companies. Its name is somewhat fitting considering how it can lead to continuous watercooler like discussions, if you let it.
[3] P2 Theme is a WordPress theme that allows teams to quickly makes new posts from the site itself and have threaded discussions on these posts. We have hundreds of these at Automattic including team based P2s, domain based P2s (eg. software testing), fun/social P2s (eg. cute baby/cat photos) and company wide announcement P2s.
[4] The Field Guide is an internal WordPress pages-only site running a custom theme which adds a few things like categorizing pages and displays contributors in the page itself.
[5] The Reader is an awesome tool that allows us to read both P2 posts as well as any blog posts from or any other site with RSS feeds. You can use it too.

13 replies on “How we Communicate at Automattic”

I have a theory that within a company, either everybody needs to work remotely (distributed) or everybody needs to be in the same office. When everyone’s distributed, you figure out ways to keep communication flowing. When it’s only partially distributed, I think the natural tendency is to optimize for in-person and realtime electronic communication, and those not in the office suffer. What are your thoughts around this?

“Even during my hiring period, all my communication was asynchronous.”

I’m really fascinated by that. But just to clarify: your interviewing with Automattic was literally all asynchronous text, with no in-person meetings or videoconferences? Or just mostly async? Either way, I’d love to hear more reflections from you at some point about how and why that works well, compared with more “traditional” technical interviews.

Yes, my entire interview process, my trial and my final interview with the CEO was all done via text chat asynchronously. The interview took about 10 days, my trial took a couple of months, and my final interview was about 7 hours.
As described in my article, that’s the way the company works so why do it different for hiring?

The benefit for candidates is you know what you’re in for, and the benefit for Automattic is they can see how someone communicates in writing is a big factor in their success for their role.

The first time I ‘spoke’ to someone in my team or my team lead was a month or so after I started full time when I attended an all company meeting in the USA.

I believe this approach also removes some cognitive bias from the hiring process, since you never see someone, you can only judge them based upon merit, that is how good they are for the job.

Leave a Reply

Your email address will not be published.