The more often they go, the more likely you encounter them.

True, but if you encounter one daily shopper, two weekly shoppers and twenty monthly shoppers, it will skew the results. Same if the numbers are reversed.

But you won't, the opposite will happen; you'll proportionally encounter way more daily shoppers than infrequent shoppers, for the simple tautological reason that the daily shoppers are going more often.

This is a sampling bias, and it's the exact same problem as if you ran a survey trying to ascertain how many people have landlines ... by calling random phone numbers.

I don’t get your point. Maybe we agree. If in one city there are 1000 30-year-old daily-shopers, 1000 50-year-old weekly-shoppers and 1000 70-year-old monthly-shoppers it will not be a good idea to estimate the age of the population by sampling people who enters the store. The average age is 50, not 34.

Most people in NYC are almost certainly not shopping monthly because of storage constraints. Even bi-weekly shopping is likely to exhaust storage space for city-dwellers in small apartments.