Exactly. I have alpha/beta testers clamoring for new features, and I have them write every single one of them down in a list. I have them reporting bugs, and I have them write every single one of them down in a list. And then I separate that list into "showstoppers" as number one priority, "user-experience" as number two, administrator niceties as number three, and legitimate, completely new features as number four.
Showstoppers means users stop using the product. Either the site is down, core functionality isn't working, it's too slow, etc. If these aren't addressed, there's no point in offering the product. Users are unforgiving, and in most cases you only have one shot.
User experience is making things nicer. Text is too big. Fonts aren't consistent. Page redirects don't go where you expect. Colors are wrong. Minor things that decrease the value the customers are getting translates into major impacts to your pocketbook. The design and finish should be as close to perfect as you can get it.
Admin nice-to-haves are things that end users won't ever see. My admin panel is atrociously designed. But only me and a handful of other people ever see it, so who cares?
Legitimately new features I actually prioritize last. Not because I don't want to offer more, but if the above three things aren't making people happy, more features won't help. And for every feature that gets added, you have to address 1-3 all over again.
What a lot of testers and developers get wrong, I think, is assuming priority #4 comes first. Numbers 3 and 4 may be able to be reversed in some situations or during a sprint, but end-user experience is always priority numbers 1 and 2. Developer/admin experience is lower down the list than a lot of developers/admins might want to believe.