Here is my answer:
First of all, the answer depends on how you would like to maximize your way. To be honest, I believe this is open-ended question because there is no best optimized $p$. In other words, what would be your confidence interval? Some likes 99%, some other 95% not 100 people to go:
For 95%
$p^{100}=0.05$, which makes our $p=0.97$
For 99%
$p^{100}=0.01$, which makes our $p=0.955$ which is practically the same answer as Gareth's answer.
and some others (like me), I would try to maximize the probability of at least 90 people go with low chance 100 will go:
maximize $C(100,99)(1-p)\times p^{99}+C(100,98)(1-p)^{2}\times p^{99}+...+C(100,90)(1-p)^{90}\times p^{90}$
which makes the result as
$p=0.9543$ with $0.93\%$ 100 people will go.
If I would try to maximize the probability of at least 95 people will go with low chance 100 people go:
$p=0.9736$ with $7 \%$ 100 people will go.
As I said, it is preference and I believe there is no exact solution for this problem.