Assume that you have a beta prior Beta$(\alpha,\beta)$ over some unknown probability $p$. I know $p$ but you do not.
Now assume that you draw $n$ samples are drawn from a Bernoulli distribution with parameter $p$, that is, $x_i\sim$Bernoulli$(p)$, $i=1,\ldots,n$. The samples are ***hidden from me***, I only know that $n$ samples were drawn.
I know that your posterior is a beta distribution, but since I don't know the samples (only the counts) I don't know what your posterior is. What is my best guess of your posterior on the probability parameter $p$?