Slack Is Where Engineering Knowledge Goes to Die
By Fahad Ijaz · · 7 min read
Somewhere in your Slack workspace, there's a message from six months ago where your lead backend engineer spent 20 minutes explaining exactly how your billing reconciliation works. It was a thorough, accurate answer with edge cases and caveats. Nobody will ever find it again. The person who asked the question moved on to the next task. The knowledge is now entombed in a thread buried under thousands of messages. When the next person has the same question, your lead engineer will answer it again from scratch.
Chat Is Optimised for Speed, Not Retrieval
Slack is a communication tool, not a knowledge management tool. It's optimised for fast, informal exchange, exactly the opposite of what knowledge preservation requires. Messages are ephemeral by design. Threads disappear behind newer conversations. Search is keyword-based and returns dozens of irrelevant results for every useful one. Paid plans retain messages, but retention isn't the same as accessibility. Having the message in your Slack history is meaningless if nobody can find it when they need it.
The Same Question, Answered Five Times
Ask any senior engineer how often they answer the same question from different people. The typical answer is 'constantly.' 'How do I set up the local dev environment?', 'What's the process for deploying to staging?', 'How does our auth flow handle token refresh?' Each answer takes 10–20 minutes and is seen by exactly one person. Over a year, senior engineers spend hundreds of hours on repeat knowledge transfer, and every answer vanishes into Slack's timeline, never to be reused.
Why Pinned Messages and Saved Items Don't Work
Teams try to solve this with pinned messages, bookmarks, and dedicated channels like #engineering-knowledge. These work for about two weeks. Then the pinned messages become outdated, the bookmarks pile up without organisation, and the knowledge channel becomes a graveyard that nobody checks before asking their question in #engineering. The fundamental problem isn't discipline. Slack wasn't designed to be a knowledge base, and no amount of workflow hacks can change that.
The Knowledge Transfer That Happens in DMs
The most damaging form of Slack-based knowledge transfer happens in direct messages. When engineers answer questions in DMs, the knowledge is visible to exactly two people. When one of them leaves the company, the knowledge is visible to exactly one person. DM-based knowledge transfer creates invisible single points of failure that don't show up until someone quits and the team discovers a critical system nobody else understands.
Moving from Chat-Based to Code-Based Knowledge
The solution isn't 'use Slack less.' It's making codebase knowledge accessible without needing Slack as the intermediary. When your code is indexed, searchable, and queryable in plain English, the PM who would have asked in #product-engineering can get an instant answer with citations. The new hire who would have DMed a senior engineer can explore the codebase independently. The knowledge lives in the code and the tool that indexes it, not in a chat message that disappears behind tomorrow's standup notes.