The trials must be dependent.

There must be more than two possible outcomes per trial.

The probability of success $p$ must change with each trial.

The number of trials $n$ must be fixed in advance.