"hedging" a review

After a few posts, I notice that I have tended to think in terms of grade ranges rather than a definitive grade for a restaurant now. While this would probably never get me a job as a food reviewer, I guess I am trying to extrapolate based on a limited amount of data. A "real" review would naturally incorporate the benefit of many visits on different days (and possibly different times) to get a comprehensive picture of the place.
As an aside, this would be an interesting optimization problem. What are the variables for such an experiment (the obvious - day, time (how busy they are), size of party etc.). Would there be a redundant visit too?
As far as this blog goes, I am trying to stick to reviewing a place only after we've visited it at least twice. Of course there are always outliers - when one visit was terrible enough not to want to go back (or when the place is sufficiently expensive to not have had an "opportunity" to revisit).