I think measuring success is relatively easy - you have GA, Tableau and similar tools to measure web traffic, purchases, etc. I think the most difficult thing is to define success ahead of a product launch. At my current company, we had large amounts of historical data from previous product launches such as page visits, page visits from the specific channels, paid vs organic traffic, revenue from every channel, etc. So we built a forecasting model that predicted the revenue forecast for Day 1, Day 30 and lifetime (1-year) for every product launch. The model was pretty accurate in predicting this, mainly because we had a large amount of historical data. You could get even more sophisticated in terms of adding external variables to this forecasting model - e.g.seasonaility, marketing events, adding new channels, etc.
Although it gets a little tricky when you don't have historical data and are launching a completely new product. But you can still come up with a model which predicts some quantifiable goal (revenue, visits, etc), make assumptions based on your research.