It is well-known that it is problematic to stop an A/B test before we obtain a required number of samples. This website "A/B Testing Rigorously (without losing your job)" explains how to stop safely an A/B test in the middle of the test.
At the website we define maximum cumulative decision limit by
$$\frac{pn}{n+10000}.$$
Here $p$ is the significance level and $n$ is the number of "conversion".
My question is: where does the formula come from?