You Probably Do Not Need Team Chat, but You Do Need Written Updates

A project slips. The retrospective says communication broke down. Someone proposes a chat tool, because chat is what communication looks like in software, and six weeks later the team has a channel for every project, four hundred unread messages, and the same project slipping.

The tool was not the problem. The problem was that decisions were being made in conversation and never written down, so every decision had to be made again a week later by whoever still remembered it. Adding a faster place to have unwritten conversations makes that worse, not better.

What chat is genuinely good at

Three jobs, and they are narrower than the category suggests. Unblocking someone right now. Presence, meaning the low-grade sense that colleagues exist and are working on something. Incident response, where the entire value is latency.

What those three share is that the information expires. A question answered in four minutes is worth something. The same question answered in four hours is a different question, because the person asking has either moved on or guessed, and the guess is now in the code.

Everything else a small team does — planning, reviewing, deciding, announcing — is worse in chat. Not unusable. Worse.

The arithmetic of an always-on channel

Take an eight-person team. Suppose each person glances at the main channel forty times a day, which is unremarkable for anyone with the app on a laptop and a phone, and each glance costs ninety seconds including the read and the return to the task.

That is one hour per person per day, eight hours across the team, or a full working day of attention spent on a tool that in most teams produces two or three actual decisions.

The ninety seconds understates it. The expensive part is the resumption lag: getting back to the state of mind you were in before the interruption takes minutes, not seconds. If only a fifth of those glances land in the middle of real work, and each costs eleven minutes of recovery, you have added roughly ninety more minutes per person per day. Treat the specific figures as illustrative — the shape is the point. A default-on channel taxes every person in the team, every day, whether or not anything is happening in it.

The counter-argument is that chat replaces email and meetings, so the comparison should be against those. Fair. But it usually does not replace them. It sits on top, and the meetings survive because chat is a bad place to decide anything.

What async written updates actually cost

A written update is not a status report. A status report tells people what you did. An update names what changed, what is stuck, and what needs a decision from someone else, which makes it a management artifact rather than a diary entry.

The cost is discipline, and it is paid in three places. The author spends thirty to forty-five minutes writing it properly. Each reader spends ten minutes reading it, which for eight people is over an hour of collective attention. And someone has to enforce the rhythm when the third week gets busy and the update quietly stops.

That last one is the whole failure mode. Async updates do not fail because people write badly. They fail because nobody is required to read, and unread writing stops being written within a month.

Compare the totals for a week: roughly two hours of team attention for a written update, against the eight-plus hours the channel was taking. The written version is cheaper by a wide margin, but only if it is actually read, and reading is a norm you have to install rather than a feature you can buy.

The three formats, honestly compared

Read the last row carefully, because it is the only row that predicts outcomes. Every format fails in its own way, and the failure you can tolerate depends on what your team is bad at. A team that decides quickly and forgets slowly should not be adding chat. A team that cannot get anyone in a room at the same time should not be adding meetings.

The format that survives a real week

Four headings, two hundred words, one author. Shipped, stuck, decided, needs a decision by Friday.

Shipped is one line per thing that changed for someone outside the team, not a commit log. Stuck is where the author names the blocker and the person who can remove it. Decided is the record: what was decided, by whom, and what it rules out. Needs a decision by Friday is the only part that generates work for other people, and it should be a maximum of three items with a named owner on each.

The constraints matter more than the template. Posted in the same place at the same time every week, written by one person rather than assembled by committee, and no discussion inside the document. Discussion goes in a single thread next to it, and anything that becomes a decision gets edited back into the document. Otherwise the update becomes a second chat channel with worse formatting.

The Friday deadline is the part teams skip. An update that only reports status is a newsletter. An update that names a decision and a date is the thing that keeps a small team from re-litigating the same argument in month three.

When you should actually buy the chat tool

Four signals, in rough order of how much they justify the spend. You run customer-facing incidents and a four-minute response changes the outcome. You have people in timezones where the written update arrives while everyone else is asleep and the latency genuinely costs money. You are past roughly a dozen people, at which point presence stops being optional because nobody knows who is working on what. You have contractors who need a room to ask questions in without being copied on email threads.

Below all four, the honest answer is that a document tool and a weekly habit will outperform a channel, and cost less. Notion, Craft, or a folder of plain files will all do it. Pick whichever your team will open without being reminded, keep it boring, and put the update at a URL people bookmark.

If you do buy, the tier structure is worth reading rather than the feature list. Free tiers usually cap message history, which means the free tier is the one that deletes your institutional memory after ninety days. Paid tiers are per active user and buy history, integrations, and guest access. The enterprise tier buys compliance and retention, which is a legal requirement rather than a communication improvement, and almost no small team has it.

The rule: chat for anything that decays in an hour, documents for anything you will need to read in a month. If a decision is not written down by Friday, it did not happen, and you will be having the same conversation again with a different tool.