Same here, Denis, same here. I don't understand why you don't like creating dummy tokens...
Your approach is relatively quick, I agree. And I understand it better now. I wouldn't say it is necessarily the "quickest" approach.
I normally prefer not to delete any data and rather create a few extra dummy tokens, then exporting a file, deleting the responses from the survey, open it, do the data entry and then import the old data. I would personally prefer to go via some dummy tokens that are also created very quickly. I think there is even a function in LS now, isn't it? And if not, you create a quick CSV file, upload it, done.
I guess it is just some kind of personal preference and "workflow" thing.