When the same door has three different labels, do you standardise or let each page phrase it?

I audited my own navigation this week and found the same destination reachable under three different names. In a banner near the top it's a hard technical label. In the footer it's warm and studio-ish. In the header it's the word people expect when there's an account involved.

None of that was a decision. Each one got written in the context of the page it sits on, months apart, and each is arguably right where it stands. The technical label is honest about what the thing actually is. The warm one is more inviting to somebody who hasn't committed to anything. The account-shaped one is the least ambiguous if you already have a login.

What I looked at:

Standardise on one label everywhere. Cheapest to do, and it makes the analytics readable — right now I cannot tell you which phrasing gets clicked, because they fire as three separate events with three different names. Cost: whichever word wins will be wrong for at least one of the three contexts.

Keep them contextual and rename the events so they aggregate. Preserves the copy, fixes the measurement. Cost: I'd be measuring the total and still learning nothing about which word did the work.

Cut two of the three entry points. Fewer doors, cleaner story. Cost: the two I'd cut sit on the pages a stranger is most likely to land on first.

I went with the second, mostly because it's reversible and the other two aren't cheap to undo.

The part I can't measure, and the reason I'm posting: for something nobody has heard of yet, does inconsistent button copy actually cost you anything? My instinct is that a stranger never holds three pages in their head at once, so each label only has to work where it stands. But I've read enough people arguing the opposite that I don't trust the instinct. Has anyone seen a difference either way?