Skip to content

An argument for threading in Slack

Hot take: Slack workspaces where users don't thread are doomed to die

Now that we are in a remote world, more of us are having to rely more heavily on chat-based communication, especially now that we are finding out how draining it is to be on Zoom calls all day.

My thoughts on this topic stemmed from the fact that my work team moved from SMS messages, to Discord, and now finally to Slack during COVID. I have never liked Slack before simply because it kept bombarding me with notifications and I found it too distracting. I was hesitant about bringing our team to Slack for this reason, until I found a Slack workspace that was very good about threading all their messages (spoiler: It's beautiful people at The Archive). To clarify, threading means that all subsequent replies and conversations should be tucked away in the thread of the original post, rather than replying in the main channel. It was an eye-opener for me to see Slack used so effectively, and this became a major impetus for us to move from Discord to Slack.

⭐️  Why we should be threading: There is more alignment between what is important/relevant, and what deserves my attention, in a way that a flat chat structure doesn't give me.

I'm convinced that threading should be used everywhere, that other chat platforms should implement it in a similar way Slack does, AND that people should learn WHY it's important to thread. The basis for my argument is founded on principles of cognitive load, distraction, and my empirical observations of chat-based communication habits.

Premise 1: Not everything you write is important

In chat based communication, we will sometimes reply with things like:

  • "Congrats!"
  • "I'm so sorry to hear"
  • "Oh look at that puppy who's a good boi"
  • "Looks good to me"
  • "ok sent"
  • jokes

Emoji reactions should be able to replace some of this "fluff", but communicating with words is irreplaceable.

Premise 2: Not everything you write HAS to be important.

So you might ask, why not just reduce all the fluff all together? After all, it would seem that the natural solution is to ban all of the types of messages in Premise 1, but that would create a very stiff atmosphere. Natural communication has interstitial content that not only define the tone, but also adds texture, to the conversation. Adding jokes, thoughts, feelings and other small concerns may be irrelevant, but they still communicate other dimensions around the content. What lie between the lines are equally as important as the lines themselves.

Censoring speech to only what is "important" also makes people feel inhibited from fully expressing their thoughts, which is the point of having a chat-based communication platform in the first place. So, while not everything is important or relevant, not everything has to be, either. We should encourage others to express whatever they want to express.

Premise 3: Not every message applies to you

Even if you were able to carve out a channel that is most relevant to your interests or job function, chances are, most topics within the channel still aren't directly relevant to you (but in a bad slack community you're going to get pinged about it all the time).

There is a difference between situationally relevant and functionally relevant content. Situationally relevant content is a "good to know" but not something urgent enough for you at that moment. Functionally relevant content means that it is meant to directly influence or guide you to take action (perform a function) on that information, urgently.

In Eisenhower matrix terms, sometimes the same content is urgent to someone else but important-not-urgent to someone else.

Premise 4: Your attention is valuable

I don't need to rehash what books and research have already said about why your attention is valuable.  If you are dependent on Slack for work, chances are, your job function falls under what is called "knowledge work" where performance requires an intensity of focus, which is thwarted by interruptions, such as chat notifications.

Dr. Sophie Leroy (then at U Minnesota) is best known for this work, where she suggests that when people switch tasks, they experience an attention residue--a "persistence of cognitive activity about a Task A even though one has stopped working on Task A and currently performs Task B."

(Highly recommend you read Cal Newport's Deep Work to get an understanding of why deep work is so important, and to get a practicum on how to incorporate it into your work.)

If you have a flat chat structure (e.g., no threading), then you either will be bombarded by all messages (see Premise 1-3), or none at all.

Premise 5: When there is no threading, every message costs the amount of same attention

In a no-thread (flat-chat) world, you would get notifications for all messages, regardless of importance. Not only are you losing focus, but you are forcing everyone else to lose focus, whether it be a constant buzzing on their phones or annoying banner alerts.

It is for this precise reason that people have started to mute ALL notifications, which is sub-optimal because it means that nobody is engaged with the workspace in a timely manner. When everyone mutes notifications, it creates even more bad habits when people demand attention, and start abusing @channel for every message because nobody was replying, which then creates even less engagement when everyone blocks the Slack app from sending notifications at all. And this is how a Slack community dies.

Slack was created to be what email was not: faster communication, bereft of signatures and "Hi" and "Sincerely" book-ending the messages, so faster communication. But speed begot irrelevance. The ease of posting messages means that Slack communities are just full of trash now, so they just start turning off notifications, which then sends the community into a devaluation spiral to become yet another annoying inbox.

Solution, and aligning relevancy and attention

Premise 1 suggests that it is natural that we not all information is important to us, and Premise 2 suggests that we shouldn't have to ban the non-important information because it still serves some purpose. Premise 3 goes further and suggests that even in a perfect world where everything written is important, a lot of things are still irrelevant, and if they are irrelevant but pestering you with notifications, then they are sapping away a critical mental resource (Premises 4, 5).

We have to fix this by using tools in combination with practice.

Channels was the original solution to the problem of aligning relevancy and attention, but it doesn't go far enough, as a lot of channel-chatter is neither important nor relevant. Emoji reactions goes a little bit further by not alerting you that someone has acknowledged your contribution.

But so far, threading is the best innovation to solve this problem:

  • You can more easily get a birds-eye view of all conversations in the channel by making only the conversation-starting message visible in the channel. Seeing yesterday's conversation doesn't require scrolling up 500 lines.
  • By tucking the replies away in threads, it actually encourages MORE conversation because people would feel less like they are imposing on other people's attention.
  • It is much easier to have more conversations in parallel. With flat-chat, you de facto shut down the previous conversation whenever you try to switch topics.
  • Threading eliminates Slack fatigue caused by getting too many notifications (which is why people turn off their notifications)

And the math also seems to check out. In my workspaces, the average Slack thread is somewhere between 10-20 messages long, with some upper tail as far as 90 messages in a single thread! If the average Slack thread were 20 messages long, that would mean that we are receiving 95% fewer notifications than we normally would, while still being able to get our work done. Work output would increase with fewer notifications.

A mental model for threads

Photo by Louis Hansel @shotsoflouis / Unsplash

A workspace where users don't thread is like being at a cocktail party where everyone in the room is the same volume to you, and you are the same volume to everyone else. There is no hiding, and there is no talking over each other. It's hard to have a conversation unless you stop everyone in the room to have that conversation with you. Switching topics is hard because you de facto shut down the previous topic in order to start the new one.

Conversely, a workspace where users thread is like being at a cocktail party where you can walk around and sample different conversations. You are invited join in any conversation, but you can also ignore any conversation by default. These conversations are more lively and more vulnerable because you don't have to feel like you're broadcasting a speech. Most importantly, you also don't have to feel like you are stuck listening in a conversation you don't care about.

I find this mental model especially fitting because at a cocktail party, there is a sense of decorum that you have to respect, namely not being crass and flooding distracting from other people's conversations. Flooding a Slack channel with messages that should be neatly tucked away in a thread is like talking really really loudly. By the time you're done, you make it very cumbersome for everyone else to scroll up to find past conversations. In the cocktail party analogy, this is roughly the same as every other guest losing track of their conversation ("what we were talking about again?").

Also, abusing the @channel is like making announcements on the mic--you should be really conscious about what you're stealing the mic for.

If you took anything away from this mental model, it's that you should thread because it is courteous and respect others' attention.


I wrote this because I am now a threading evangelist, but it's not enough to tell people to thread, they have to understand why they should adopt this standard. I wrote this for all my fellow frustrated threaders out there living amongst non-threaders. Have patience, they just haven't seen the light. I hope my argument here can thoroughly explain the perspective of the practice and convert them.

To some degree, I do think it's Slack's fault for not creating a better design that pushes for threading. One crude design solution is to enable "slow mode" in the main channels via the Slackbot (the Slo app offers this, but only for private channels). There are several ways this could happen: 1) No single person can send a message in channel more than once within X minutes, or 2) no messages can be sent less than X minutes apart. This could force people to reply in threads (like if there was a nudge?).

At some point in the future I'll write about the best practices of threading, and how Slack can design a way to put threading first. If I haven't written it yet, hold me accountable and bother me about it.