How to break down information silos

Information silos are one of those problems that every organisation acknowledges and almost none successfully fix. The standard diagnosis is cultural: teams aren't collaborating enough, people are hoarding information, departments don't talk to each other. The standard prescription follows logically: encourage collaboration, reward sharing, create cross-functional teams, hold more interdepartmental meetings.

These interventions produce modest results because they're treating a structural problem as a cultural one. Silos don't form because people choose to hoard information. They form because each team picks the best tool for their work, and those tools don't connect.

How silos actually form

Nobody decides to create an information silo. They emerge naturally from reasonable decisions made independently.

Sales adopts a CRM because CRMs are built for managing pipelines and client relationships. Engineering uses GitHub because GitHub is built for code. Marketing uses Google Drive because it's built for collaborative documents. Design uses Figma because it's built for design work. Product uses Linear or Jira because they're built for project tracking. Finance uses spreadsheets because spreadsheets are built for financial modelling.

Each of these decisions is individually rational. Each tool is good at its specific job. The problem is that none of them know the others exist. The client context in the CRM is invisible to the engineer who's building the feature that client requested. The design decisions in Figma are invisible to the product manager tracking the project in Linear. The market research in Google Drive is invisible to the sales team who could use it in their pitch.

The silo formed without anyone choosing to create it, which is why telling people to share more doesn't fix it. The sharing problem is a plumbing problem, not a behaviour problem.

Why cultural solutions fail

"Encourage sharing" fails because sharing across silos requires extra effort. An engineer who discovers something useful for the sales team would have to actively decide to share it, figure out where the sales team keeps their information, navigate an unfamiliar tool, and post it in a way that makes sense to people with a different context. The friction is high enough that it almost never happens, regardless of how strongly the culture encourages it.

"Cross-functional meetings" fail because they add synchronous overhead to a problem that should be solved asynchronously. The monthly cross-team sync produces a brief period of alignment that decays within days as each team returns to its own tools and its own context.

"Centralise everything into one tool" fails because no single tool is good enough at everything. Forcing engineers to use Notion instead of GitHub, or forcing sales to use Confluence instead of their CRM, degrades the experience for both teams and produces resistance that undermines the entire initiative.

The connective layer

The approach that works is a layer that connects existing tools without replacing them. Each team keeps the tool that works best for their specific work. The layer indexes the content across all of them and makes it searchable from one place.

An engineer searching for context about a client request finds the CRM entry, the product brief, and the Slack conversation where the request was discussed, all from one search. A product manager searching for the design rationale finds the Figma comments, the design review meeting notes, and the relevant Slack thread. The information stays where it was created. The search spans everything.

Fabric's connections work this way: Google Drive, Slack, GitHub, Gmail, Linear, Figma, HubSpot, Notion, and dozens of other tools feed into one searchable library. The silo breaks because the search layer spans the boundaries that the individual tools create.

Through MCP, this unified knowledge layer also becomes available to any AI agent the team uses. The agent that can search across all your tools, understanding context from every department, is qualitatively more useful than one that can only see the data in whatever tool it's natively integrated with.

The practical path

Breaking silos doesn't require a migration project or an organisational restructuring. It requires connecting the tools you already use to a shared search layer.

Start with the highest-friction boundaries. Which cross-team information gaps cause the most meetings, the most duplicated work, the most delayed decisions? Connect those tools first.

Don't ask people to change their workflow. The engineer continues using GitHub. The designer continues using Figma. The product manager continues using Linear. The change is that all of them can now search across each other's tools when they need context, without navigating unfamiliar interfaces.

Let search do the work that meetings currently do. Many cross-team meetings exist to transfer context that could be found through search if the tools were connected. As the search layer covers more sources, these meetings become optional.

Track the reduction in cross-team questions. The measure of success is the frequency of "does anyone know...?" messages in Slack. When people can find the answer themselves, the question doesn't get asked, and the person who would have answered doesn't get interrupted.

Frequently asked questions

How is this different from enterprise search tools like Glean? The principle is similar: connect tools and make them searchable. The difference is in who it's designed for and what it connects to. Fabric is built for teams from 5 to 500, not just enterprises, and it combines the search layer with knowledge management features (notes, docs, AI assistant, self-writing docs) that make the discovered information actionable, not just findable.

Will this work if different teams have different security requirements? Yes. The search layer respects the permissions of the underlying tools. Content that's restricted in the source tool remains restricted in the search results. You can also configure which sources are included and which are excluded, giving each team control over what's searchable across the organisation.

How long does it take to see results? Connecting tools typically takes hours, not weeks. The search layer begins producing results as soon as the first sources are indexed. The reduction in cross-team friction is usually noticeable within the first month as people discover they can find answers they previously had to ask for.

Related reading: The cost of scattered knowledge, The knowledge scaling problem, Information silos are the default, The app sprawl problem. Related pages: Connections, One search, Find anything.