Why your wiki doesn't stop Slack pings

Every technical B2B company has a wiki graveyard. Confluence pages titled "Architecture Decision Records" with three entries from 2021. Notion workspaces someone started and abandoned. A Google Doc called "Source of Truth" that nobody trusts.

Then Slack rings. Same question. Same senior person. Again.

The wiki did not fail because people are lazy. It failed because decisions move faster than documentation. What is a company brain? walks through the alternative.

The interruption tax

Ask any eng lead at a 100-person company how many hours per week they spend re-explaining past decisions. Vendor choices. Service boundaries. Why billing is separate from auth. What sales can promise in a pitch.

The answers are usually somewhere. A thread from March. A ticket someone closed. One person's head. They were never pulled together because consolidation was homework, and homework loses to shipping.

Enterprise search was supposed to fix this. Connect every tool, ask a question, get an answer. Search retrieves what was written. The "why we chose Stripe over Braintree in Q3" conversation may never have been written anywhere except Slack, and even there it sits under reactions and side threads.

That gap is what people mean when they talk about institutional knowledge walking out the door. The docs are fine. The judgment never made it into a doc.

Why mandates do not work

Top-down "update the wiki" programs die for predictable reasons. Documentation happens after the decision, when context is already fading. Individual contributors are measured on shipping, not on graph density. Wikis want polished pages. Decisions are often one paragraph in #finance. A wiki page does not link to the Slack thread where finance and legal actually agreed.

One pattern shows up repeatedly: the quarterly documentation sprint. A VP declares a week for "knowledge capture." Twenty pages appear. Traffic analytics show single-digit views within thirty days. The senior eng still gets pinged on Monday because nobody remembered the sprint happened, or the page title did not match how people ask the question.

You cannot mandate your way out of tacit knowledge. Capture has to happen at decision time with friction measured in seconds, not hours.

What works instead

Teams that reduce the interruption tax tend to share a few habits. None of them involve a bigger wiki template.

Capture when the decision is fresh. A ten-second prompt right after a thread resolves beats a quarterly documentation sprint. The people who were in the room are still there. The constraints are still in working memory. When Slack decisions never become docs walks through the thread lifecycle and why timing matters.

Human confirm before save. Surveillance breaks trust. Explicit confirm builds it. Nothing lands in the Company Brain unless someone approves it. That is also how teams avoid polluting memory with half-formed takes and jokes taken out of context.

Scoped retrieval, not one pile. An Operator escalation path and a Portfolios onboarding rule should not share the same retrieval space. How context groups keep answers scoped explains how Ignius handles this with Operator, Portfolios, Partner, and Employee groups.

Citations, not vibes. Answers should link back to Office threads, GitHub PRs, or Xero invoices, not a model paraphrase with no source. If someone cannot verify where an answer came from, they will ping the senior anyway.

That is the design bet behind Ignius: not another wiki, not another search box, but a Company Brain that grows when people save what mattered once.

Search helped. It did not close the loop.

A lot of teams bought enterprise search expecting the pings to stop. Relevance improved. Doc retrieval got faster. The "why" questions kept coming because the answers were never written down in the first place.

Glean searches memory. Ignius creates it. explains why search and memory creation solve different problems. For a side-by-side breakdown, see Ignius vs Glean.

The test you can run this week

Pick one question your team asked twice in the last month. Try to answer it from the wiki and from Slack search. Then ask whether the decision is retrievable: who decided, when, and under what constraints.

If the answer is no, the team does not have a search problem. It has a memory creation problem.

Read how Ignius thinks about institutional knowledge or see Company Brain for the team layer.