I would guess that it would not be particularly uncommon for someone to provide a rating one day, and then adjust it if asked the same question a little later.
The nice thing is that if you are collecting enough responses, any odd circumstances around a single rating here or there become minimized and hopefully irrelevant as the sample size increases.
If you have a small group you are sampling I could see that causing some concern. But what is really the difference between a 6 and 7 rating? The customer is still in the same ball park. If instead, your customer gave a 6 rating and then later stated that they meant to give a 0, or a 10 you might have a problem worth following up with.
