Never do a large-scale intervention without building in an evaluation. 25% of your intervention budget should be evaluation because you do not know that your stupid intervention, which you think will do what it you think it will do, that's just a guess. It's a guess, and it could go wildly wrong ten ways you don't predict.