Comfort Isn’t Strategy

Every time I suggest moving away from folders in SharePoint Online, someone laughs.

And I understand why.

Folders feel organized. They feel grown up. They feel like the responsible way to structure information. The digital version of labeled drawers and neat rows.

If you grew up in file shares, folders were the system. They mirrored filing cabinets. They created hierarchy. They gave you something concrete to click into when you were not sure where to start.

They feel mature.

But comfort, even grown up comfort, is not strategy.

A shared SharePoint library is not a filing cabinet. It is collaborative. It is indexed by search. It feeds Copilot. It supports knowledge experiences that depend on structure being interpretable, not just clickable.

That difference matters more than we want to admit.

Folders can work beautifully in your personal OneDrive. If you are the only one navigating that space and you understand your own logic, number things however you want. Build your maze. Save FINAL_v8_THISONE_FOR_REAL.

That is your sandbox.

The problem is when that same model becomes the default in shared environments.

What feels logical to the person who built the structure often feels confusing to everyone else, especially six months later when context has faded and teams have shifted. What once felt organized slowly becomes something people navigate by memory instead of clarity.

That friction is subtle, but it adds up.

Folders assume everyone approaches information the same way.

We do not.

Some people remember the client name. Others remember the year. Some think in terms of department. Others think in terms of project status. Some remember a single keyword buried in a title. When folders dominate the structure, they favor one mental model and quietly frustrate the rest. If your brain does not match that model, you spend more time clicking and less time working.

Metadata changes that.

Instead of forcing one path, it creates multiple entry points. A document can carry attributes like Budget, North America, FY24, Approved, Cost Center 1142. Now it can surface based on how someone naturally approaches the question. Region. Year. Status. Department. The structure adapts to the person instead of forcing the person to adapt to the structure.

That is not bureaucracy. That is usability.

And in the Copilot era, it becomes foundational.

Copilot does not browse folders the way we do. It looks for patterns and signals. It interprets structure through attributes and relationships. When content is buried in deeply nested hierarchies with inconsistent naming, the system has very little clear signal to rely on. It can infer meaning, but inference introduces variability.

Sometimes the response feels accurate. Other times it feels almost right. Confident. Polished. Slightly off.

That almost is where trust starts to erode.

A user asks, “Show me the latest approved North America budget.” The answer is not quite right. They try again. Still uncertain. Eventually they stop asking and go back to emailing attachments or saving files locally.

Nothing dramatic happens.

Adoption just stalls.

This is why the folder conversation matters.

It also shows up in tooling. In my experience, SharePoint’s Knowledge Agent capabilities do not work in foldered libraries. When structure is built primarily around nested folders, dynamic organization and retrieval become limited or inconsistent. The tool is not broken. The architecture simply was not designed for how the tool now operates.

Folders are not inherently wrong. They are familiar.

And familiarity is powerful.

Blockbuster Video worked beautifully for the era it served. It was predictable and comfortable until the environment changed.

We are in one of those shifts now.

If we want Copilot to feel reliable, not just impressive in a demo, structure cannot be an afterthought. That does not mean deleting every folder tomorrow. It means recognizing that a shared, AI powered workspace deserves more intentional design than a digital filing cabinet.

Start small. Choose one meaningful library such as contracts, policies, budgets, or project deliverables. Simplify the hierarchy. Introduce a handful of well defined metadata columns grounded in the real questions people ask. Create views that allow filtering by region, year, status, or department. Let people experience how much easier it becomes to retrieve information when structure supports multiple ways of thinking.

Then test it honestly. Ask Copilot the questions your users actually ask. See what changes. Adjust.

Maybe the real question is not whether folders are good or bad.

Maybe it is whether the way we structure information today gives our people and our AI tools the best chance to succeed.

Take one library. Just one. Look at it without defending it. Ask yourself how many meaningful questions it can answer without someone already knowing exactly where to click.

If the answer is not many, that is not criticism. It is clarity.

This work rarely makes it into keynote demos. But it is the difference between Copilot feeling novel for a week and becoming something people actually trust.

If you are rethinking your SharePoint structure in light of AI, or noticing inconsistent Copilot results, this is often where the real work begins.

And it is a conversation worth having.