Nevermind 25 approvals, they do not represent the real figure as well. What rupnet shows is a ratio of approvals. If, statistically speaking, every service center has the same volume of people representing it on rupnet, we would see at what rate each center processes its cases.
But, unfortunately, rupnet had not gathered enough volume to become a fair source of information. I'm not implying that the information is completely useless, I'm rather saying that it should be allowed for an error.