I've spent nearly 30 years working in English.
I still refuse to troubleshoot a new tool in English. Not because I can't read it. Because when I'm stuck and stressed, the last thing I need is extra cognitive load.
This is not an English proficiency problem. This is a brain capacity problem.
When users are confused, frustrated, or just trying to get something done fast, they're already running at full capacity. Add a foreign language on top of that, and they don't push through. They leave.
The power user assumption
This is why "our power users are fine with English" is a dangerous assumption. Power users aren't confused. They're not stressed. They have cognitive bandwidth to spare.
Your struggling users don't.
And struggling users are exactly who your help content is for.
What happens when help content isn't there
My workaround? I tell an external AI: "Navigate me through this tool. The UI is in English — guide me in Japanese."
Not the AI assistant built into the tool. A separate AI, because the built-in assistant couldn't help me navigate the product itself.
It works. But that means the tool lost me — and another AI caught me.
"Just add an AI assistant" is not the answer
Localized help content still matters. It gets indexed by Google. It helps confused prospects find you in their own language, at the moment they need you most. And once users are inside the product, it gives them a way forward without forcing them to outsource comprehension to another tool.
If your users are already inside the product and still can't move forward, that is not just a language barrier.
That is a design failure.