You do need a bit more information. My instinct is to consider this as a binomial distribution, therefore assuming indepence etc... So what is needed for this approach is the probability that a client purchases something. You could let this be $p$(say) and then calculate the probabilites assuming this value. for example; Let $X$ be the number of clients who make a purchase with probability $p$. Then $X \sim Binomial(15, p)$
a) $P(x=0) = (1-p)^{15}$
b) $P(X =12) = \binom{15}{12} p^{12} (1-p)^{3}$
c) $P(X \le 20) = 1$ Since $20 > 15$.
You could use a Poisson model but then some rate parameter needs to be assumed and a similar method can be used as above using the Poisson distribution instead.
Overall more information is needed in order to explicitly give solutions.