There's a quiet frustration baked into this question, and it's one we recognize all too well: the gap between what people *should* do and what they *actually* do. The user has a clear rule, avoid letters like O, I, S, L, and N to prevent confusion with numbers, but the rule only works if every single person remembers it, cares about it, and applies it consistently. They don't. And so the data entry errors pile up, the corrections eat away at everyone's time, and the person who built the form becomes the de facto enforcer of a policy that was never enforced in the first place.
What stands out here is the attempted solution. The custom validation formula is a reasonable instinct, but it's incomplete, it checks for a few specific letters while missing others, and it doesn't scale with the actual problem. The real issue isn't the formula; it's that you're relying on the person typing to make a judgment call in the moment. That's a design flaw, not a discipline flaw. If the system allows an "O" to be entered, someone will enter it. The fix isn't to remind people to be careful. The fix is to make it impossible to enter the wrong thing in the first place.
This is where the conversation about tools like Kools becomes relevant, but it also reveals a deeper point: a tool only helps if it's present everywhere the work happens. If the form is copied to desktops that don't have the tool installed, you're back to square one, except now you've added a new dependency that not everyone has. That's not a solution; that's a patch on a patch. The real solution is to build the restriction into the data entry layer itself, so it travels with the file, not with the software.
The practical takeaway is simple: stop trying to police behavior after the fact. Instead, design the input field so that confusing characters are never accepted, regardless of who opens the file or where it lives. That means using a validation rule that blocks the entire set of problematic letters, not just a few, and testing it across the actual environments where people work. If a tool doesn't work everywhere, it's not a solution, it's another variable. Start with the validation logic, make it airtight, and then, if needed, explore external tools only after you've confirmed they work universally across your team's setup. The goal isn't to correct mistakes faster. It's to make the mistake impossible to make.