From economic perspective it makes sense to stay open as long as the marginal revenue is higher than marginal cost of staying open. All fixed costs are de facto sunk in the short-run. In addition, since the number of customers store gets within hour is random, what matters is that the expected marginal revenue is higher than marginal cost.
For example, if some hot-dog stand sells hot-dogs for \$10 and marginal costs for operating per hour is \$20, having a 67% chance of having only 3 customers (each having single hot-dog) per hour will be sufficient to keep the store open (assuming for the simplicity that either 3 people show up or nobody will).
In the example above you could observe hot-dog stand having no customers for 1/3 of a day and it would still make perfect sense for the stand to operate 24/7.
Above is of course just an example. Depending on parameters of the problem (marginal revenue, marginal costs etc), you can construct examples where it makes sense to keep store open even if probability of having a customers per hour is very low. For example, suppose its a jewellery store that gets marginal revenue of \$1000 per sale but has marginal costs of only \$50. Then a probability of having single sale 5% per hour (vs no sale per hour), would be sufficient to keep the store open (5% probability of one sale means you would expect to see about one customer per day).
Even if shop would be profitable enough just being open during rush hour it would be irrational not being open other hours as long as expected marginal revenue is higher than marginal costs.