A surprising number of conversion-rate-optimisation projects are really event-quality projects in a trench coat.

The short version: experiments whose results you actually trust is mostly a matter of doing the obvious thing carefully. Get the form events right server-side first, then run the experiments. Reverse the order and you will optimise against noise.

You spend a couple of weeks on instrumentation before you can run experiments. The experiments are worth more. That trade-off is the one most teams underweight when they plan the work, and it is the one that quietly determines whether the project ships on time.

If you want a longer treatment, we have a deeper post in the series. For now, the takeaway is small enough to fit on a sticky note: Most CRO projects benefit more from cleaner measurement than from another A/B test.