Say no to the good ideas

Nobody loses a year to an obviously bad idea. The dangerous ones are the plausible adjacent bets: the integration a big customer asked for, the second product line, the feature three people mentioned in the same week. Each one is defensible on its own. Together they are how a focused company becomes an unfocused one without any single decision being wrong.

The test that works

Ask what you would have to stop doing to do this properly. If the answer is "nothing, we'll fit it in", the idea has not been costed and the answer is no.