It's actually super useful if you've got a table with two columns like "sku" and "product_name", where the value in both columns has one unique value from the other column paired with it. You can group by sku, and then grab any value for product_name, since they're all the same. Using min/max/etc involves extra computations that you don't need to do.

You could group by the combination of sku+product_name, but it's kind of unwieldy. You could also normalize it to not have this issue, but when you're getting this data from someone else, there's not much you can do.