Skip to content
AppDues
Log in
Contact salesSign up

Remote & Async Communication: A Practical Team Guide

Distributed teams rarely fall apart because conversation stops. They struggle when discussion lands in the wrong place, moves at the wrong pace, or disappears without a record. This guide offers a practical set of principles for asynchronous, remote-first communication and then matches each common internal tool to the work it handles best. It is intended for anyone writing or revising a team communication handbook. The practices described here are adapted from a publicly shared internal communications guide published by another company as inspiration for remote teams; that original drew on GitLab's communication handbook. What follows describes those practices, not the internal operations of the publisher of this page. Attribution and adaptation details for this version are maintained separately by the site owner.

Seven principles that hold a distributed team together

These principles are not abstract values; they are the operating rules that make remote work possible. Each one answers a recurring question: should I interrupt, where should this conversation live, and what do I owe the next reader?

Protect focus. Asynchronous work means information arrives when the recipient can process it. Nobody is expected to answer a weekend email before Monday. Urgency is the exception, and it should be used sparingly — more things feel urgent than actually are, and over-escalating quietly erodes someone else's right to concentrate. The point of the rule is simple: most people need uninterrupted head space to do their best work.

Default to transparency. Withholding information corrodes trust, and distrust in turn breeds resentment. Being open has real costs — you may invite criticism you did not want, or raise concerns before you can address them — but a team that can absorb those costs tends to be healthier and more mature than one that cannot. The same instinct applies externally: sharing with a developer community can build affection for a brand, while undisclosed investors, customer identities, and proprietary know-how stay private.

Go public, not private. Ask questions where everyone can see them — public channels, shared docs, issue trackers, distribution lists — rather than one-to-one messages. If you need a specific person, name them in the public thread so others can still learn and contribute. When a conversation genuinely needs privacy, write up the outcome and post it publicly. A public record is searchable, and it defaults the team toward transparency.

Provide context. Any reference to a pull request, issue, commit, page, or comment should carry a link. People with low context are the ones most able to help, and they cannot help what they cannot find.

Be specific when you pull someone in. A CC or a channel add is not a request. If you need action, mention the person and state plainly what you need.

Stay human. Rules can read as robotic. On video, it is fine if a child or partner wanders into frame — that is one of the perks of remote work, and it reminds everyone you are a person rather than generated text.

Match the channel to the job. Every tool has a different level of bandwidth (how rich and interactive it is) and synchrony (how live it is). A video call is high-bandwidth and synchronous; a recorded video is asynchronous; a chat message is low-bandwidth. Default to async and high bandwidth, then work backwards only when the situation demands more immediacy.

Which channel for which job

The fastest way to make a communication guide usable is to answer one question: where does this thing live? The table below turns the guide's rules into a decision aid. Read it top to bottom and pick the first row that matches your situation.

Using chat without letting it take over

Chat works best for informal exchange. Approvals and decisions belong in a document, an issue tracker, or email — somewhere that does not scroll out of sight. A few habits keep chat healthy: prefer a public channel and mention the person you need, since private messages hide the answer from everyone else; skip private group DMs, because every member is pinged on every message, the group cannot be searched or shared, there is no subject line to recall the topic, and leaving the group wipes the history — spin up a temporary channel instead and archive it when the work is done, and note that private group DMs also tend to multiply because you cannot add people to an existing group, so people create new ones for the same topic; do not expect presence, and treat chat as asynchronous, escalating to a phone call only if something is truly an emergency; skip the greeting, since a bare hello interrupts someone without saying anything — ask the question directly, and if you need a live conversation, say so explicitly; use threads so parallel topics do not block a channel; turn on Do not disturb while focusing and disable mobile notifications outside working hours, and keep a separate emergency contact route in place — the original guide recommended this, and each organization should confirm its own route; avoid @here and @channel, which are almost never appropriate; remember email exists, because for some messages it is simply the better tool; respect out-of-office status, and do not mention someone on vacation in a high-volume channel — send them the link directly instead; leave and archive channels freely, since neither is rude and both reduce noise; and fill out your profile. Naming conventions help too: prefixes force the creator to think about a channel's purpose and make channels easy to group. The prefixes used in the original guide are: #topic- — long-term discussion of a specific subject; #feed- — mostly automated messages plus the conversation around them; #team- — team water coolers, loosely mapped to the org structure; #guild- — cross-team groups with shared skills; #fun- — random and silly; #interview- — private, for coordinating a candidate's interviews, archived afterwards; #feature- — cross-functional work on one feature, time-limited. The original guide used these prefixes, and each organization should confirm its own conventions. Pull the emergency brake when a thread runs long: long threads where nobody owns the decision are nobody's fault — people are trying to help — but they are hard to catch up on and rarely the right tool. Anyone, participant or not, can post the :noslack: emoji to stop the thread. The next step is either a document collecting the options and trade-offs, or a scheduled live discussion. Standups can help as well: a chat-driven standup bot acts as the company heartbeat and gives a quick feed of what is moving. Keep updates honest and brief rather than performative; configuration is per team. The original guide used Geekbot, and each organization should confirm its own standup tool.

Documents, wikis, and work items

Collaborative documents are the right home for decisions that need high bandwidth but not synchrony. The original guide described this typical lifecycle: the owner drafts the decision document and links out to supporting material; the owner asks a small circle of core contributors for comments or amendments, and the document is still a draft at this point; once those comments are resolved, the owner marks it for review or comments and opens it to a wider group — the cross-functional team, all of engineering, or the whole company, depending on the topic; the original guide kept documents in review for at least three days, and each organization should confirm its own review window; the owner resolves every comment, including the ones they decide not to act on; if the decision maker is someone other than the owner, they make the call and it is recorded in the document; and the document is then marked final. Wikis hold curated information that lasts. If something is temporary or needs heavy collaborative editing, use a collaborative document instead. Anything that must persist and stay visible to others belongs in the wiki, where comments, questions, and direct tags are welcome. Issue trackers and design tools. Comments on a pull request, ticket, or design file are preferred when the scope is limited to that work item. When the discussion outgrows the item, escalate it: comments about the right implementation for a ticket can live on the pull request, but if the question becomes what functionality to build, move it to the ticket, since not everyone reads pull request comments; a ticket can host comments about the right approach to a problem, but if the problem itself is in question, escalate to the product spec; and design comments can cover whether form fields are correct, but if an entire flow is missing, that needs a new ticket or a spec change.

Email and calendar habits

Email. The core weakness of email is that it hides conversations from anyone not on the original thread, and future colleagues cannot reconstruct why a decision was made. Distribution lists fix much of this because they archive every conversation. Create one for any team or initiative where important decisions are likely to happen over email, and use broader lists for announcements and proposals that need wide feedback. Other rules: one subject per email, since bundling items causes delays (someone must answer everything) or misses (one item gets forgotten), and if a separate topic emerges mid-thread, start a new thread with the right people; write a useful subject line, where RFC signals a long request for consideration, FYI signals information that may not need reading, and a bracketed deliverable with a deadline, such as [Something needed] By EOD Wednesday, makes the ask obvious; reply all; treat email as asynchronous, so a weekend message can wait for the workweek; and if an email becomes urgent, ping the person through chat or phone and reference the email, but limit the did-you-get-my-email follow-ups. Calendar. Share your calendar broadly enough that colleagues can see event details. The original guide recommended a company-wide see-all-event-details setting, and each organization should confirm its own sharing level. Mark personal appointments, one-to-one performance or evaluation meetings, and any meeting covering sensitive information as private. Sharing pays off in two ways: it supports a culture of openness, and it makes scheduling across time zones far easier, because colleagues can see recurring commitments and find a slot without waiting for a confirmation. If you block time for recurring personal tasks, mark yourself free for those blocks so others know they may book over them when nothing else fits. Event-level visibility settings let you add personal items to a work calendar when you want more privacy. Signalling availability. If your schedule is unusual, these settings matter more: put planned away time — holidays, vacation, travel, other leave — on the shared team calendar; add the same out-of-office block to your personal calendar so meetings auto-decline; set working hours in calendar settings so your off-hours are visible across time zones; and use chat's do-not-disturb mode and status to show availability.

When a call is worth it — and how to run one

Synchronous calls should be the exception. Before booking one, check whether an asynchronous route reaches the same outcome; regular one-to-ones are the standing exception. Live calls take scheduling control away from attendees, pull people out of deep work, and are awkward to arrange across a globally spread team. They are also costly at scale: only one voice can carry at a time, so a large meeting leaves much of the room idle. Decide to call when you have gone back and forth about three times over email or chat, when you are blocked and a short conversation would unblock you, or when the topic needs genuine live debate. Choose something else when the attendee list is long — a document or a recorded video reaches more people more cheaply — or when the purpose is broadcasting information, in which case record it instead. Prepare the invite: every scheduled meeting needs a video link or dial-in details, and you should link any documents or shared notes you will discuss and confirm everyone has access. Run the call well: give it an agenda that states the outcomes you want, not just the topics, since discussing a subject is not an objective and making a decision is; do as much preparation asynchronously as possible, sharing documents in advance, recording a proposal beforehand if it needs presenting, and for a retrospective collecting proposed topics ahead of time so the call is spent discussing them; record the outcomes and post them publicly, agreeing at the start if it is unclear who owns the meeting, and for recurring meetings keeping a running agenda document that also captures outcomes per item; and consider recording the call, especially when people cannot attend, and share the recording. If your recording service deletes files after a set period, download and store them somewhere durable — the original guide noted a 14-day deletion window for one tool, and each organization should confirm its own retention policy. A recording does not replace written outcomes, because nobody should have to watch thirty minutes to find one decision. Expectations on the call itself: start on time and do not wait for latecomers; use headphones, since ambient noise and echo degrade the call, and if you need a headset, check your organization's equipment policy — one person on speakers is tolerable, several is not; turn the camera on, pointed at your face, with reasonable lighting, because reading faces improves the conversation, though there are exceptions and lacking video is not a reason to skip a meeting; default to unmuted, since muting creates a small barrier to the spontaneous interruptions that lead to creative collaboration, and find a quiet space when you can; do not be afraid to interrupt, because latency makes it feel ruder than it is in person, but the questions and context interruptions supply are valuable, so be thoughtful about when, whom, and how you interrupt; and own your connection, since a stable internet connection is your responsibility — if you need help at your primary workplace, check what support your organization offers, and if problems are likely, plan to dial in by voice or move the meeting. Recorded video suits broadcasting to a wider group in a way that feels more natural and less polished than a document. It also makes calls more asynchronous when the presentation portion is shared up front, and it is excellent for product demos or showing a bug.

A worked example: one question, three channels

Suppose an engineer asks in a public channel whether a checkout screen should show a trial badge. Within an hour, six people have replied and the thread has split into pricing policy, design consistency, and implementation detail. Nobody has decided anything.

Here is how the rules above route it:

  1. Chat was the right place to raise the question, because it was public and visible. It stops being the right place once the thread passes a handful of exchanges.
  2. Someone pulls the brake with :noslack: and offers to write it up. The thread is closed.
  3. A collaborative document collects the options, the trade-offs, and a recommendation. The owner drafts it, asks two core contributors for comments, then opens it to the cross-functional group for at least three days. The decision maker records the final call in the document.
  4. The wiki receives the durable version — the policy that future hires will need.
  5. The issue tracker receives the implementation work: a ticket for the badge, and a separate ticket if the discussion revealed a missing flow.

The lesson is not that chat is bad. It is that each channel has a job, and a question that starts in one place often needs to graduate to another before it can be answered.

A checklist before you hit send

Use this before posting, emailing, or booking anything:

  • Is this public by default? Could it be a channel post or a shared document instead of a direct message?
  • Does it link to the pull request, ticket, page, or comment it refers to?
  • If I need action from a specific person, have I named them and stated the ask?
  • Is this a decision that needs a durable record, or just a quick exchange?
  • Have we already gone back and forth three times? If so, book a short call.
  • Does the meeting invite contain a link, the agenda, and the outcomes we want?
  • Who will write down what we decided, and where will it be posted?
  • Am I marking this urgent because it is, or because it is convenient for me?
Channel selection: match the tool to the job
SituationBest homeWhy
Quick question, informal exchangePublic chat channelVisible, searchable, others can chime in
Approval or decisionDocument, issue tracker, or emailChat scrolls away; decisions need a durable record
Temporary group discussionTemporary channel, archived afterwardsPrivate group DMs are noisy, unsearchable, and lose history
Decision needing wide inputCollaborative documentHigh bandwidth but asynchronous; staged review
Durable reference materialWikiWritten rarely, read often
Work item or feature requestIssue trackerRepresents a unit of work; discussion stays attached
Broadcast to many peopleRecorded video or documentReaches more people than a large call
Complex, immediate, or blockedVideo callHigh bandwidth and synchronous; use sparingly

This table restates the guide's own routing rules. It is a decision aid, not a survey of other companies' practices.

Frequently asked questions

How quickly should I expect a reply to a chat message?

Treat chat as asynchronous. There is no expectation of an instant response, because you cannot know what the other person is doing. If something genuinely requires immediate attention, escalate to a phone call.

Why avoid private group DMs?

They notify every member on every message, cannot be searched or shared, have no subject line, and lose their history when someone leaves. A temporary channel that you archive afterwards does the same job better.

Is it rude to leave a channel or archive one?

No. Leaving a channel once your question is answered keeps it from distracting you, and archiving channels aggressively keeps the workspace manageable.

How long should a decision document stay open for review?

At least three days, so everyone has a chance to comment. The owner then resolves every comment — including the ones they choose not to act on — and marks the document final. Each organization should confirm its own review window.

When should I book a video call instead of writing?

A useful rule of thumb is three rounds of back-and-forth. If you have exchanged messages three times without resolution, move to a call. Also call when you are blocked and a short conversation would unblock you. Avoid calls for large groups or pure broadcasting.

Do I have to be on camera?

Camera on is preferred because reading faces improves the conversation, but there are exceptions. Not being able to use video is not a reason to skip a meeting.

What is the emergency brake in a long chat thread?

Anyone who feels overwhelmed by a long thread — participant or not — can post the :noslack: emoji to stop it. The thread then moves to a document that collects the options and trade-offs, or to a scheduled live discussion.

Should I record meetings?

Consider it, especially when people cannot attend. Share the recording, but do not let it replace written outcomes — nobody should have to watch a full discussion to find a single decision. If your recording service deletes files after a set period, download and store them somewhere durable.