The pattern I keep running into with Copilot rollouts: the reaction from employees is almost always some version of the same comment: it's not as good as ChatGPT.

That's not really about capability. It's about comparison.

The comparison isn't fair, but it's the one people make

An employee who's spent a year or two on a personal ChatGPT account has built up months of context around their own writing style, their recurring topics, their way of framing a question. Copilot shows up on day one of an enterprise rollout with none of that history. Microsoft has started rolling out its own memory features, but a fresh deployment still starts from zero.

So the first few answers feel generic by comparison, and generic reads as worse, even when the actual capability gap is much smaller than it feels.

Stop competing on memory. Compete on data.

ChatGPT can now connect to inboxes and SharePoint too, but only as a separate integration sitting outside the Microsoft Enterprise Agreement the company already has: its own contract, its own compliance review, its own admin approval. Copilot's access to that same data comes for free with the tenant the company already governs. That's not a personalization trick, it's a structural advantage that doesn't require a new agreement to unlock.

The rollouts that land aren't the ones that wait for Copilot to feel as personal as ChatGPT. They're the ones that point people at tasks only Copilot can actually do: summarize the meeting you were just in, draft a reply referencing the email thread you're looking at, pull a number out of last quarter's report sitting in your files.

A simple test

Ask someone who's underwhelmed by Copilot to try it on one task: something tied to a real file, email or meeting they had this week, not a general question.

If the answer suddenly feels useful, the problem was never capability. It was what they were asking it to do.

"The fix isn't giving Copilot more memory. It's pointing people at the one thing memory can't do: read your actual files."

Set the right expectation from day one

Copilot was never going to win a personality contest against an account someone's spent two years building context with. Positioning it as a better ChatGPT sets it up to lose that comparison every time. Positioning it as the tool that already knows your company's data plays to the one advantage ChatGPT can't copy.