Here's a plan B based on several experiences with enterprise SaaS: Start with something that isn't awful. Get it up early. A/B test that crap out of out. Keep

Here's a plan B based on several experiences with enterprise SaaS: Start with something that isn't awful. Get it up early. A/B test that crap out of out. Keep it simple -- Only have one unit of measure (UoM) or feature differentiator at first. Having multiple points of differentiation or UoM leads to confounded experimentation. Try to add more " per's " over time :) You can take forever debating this stuff. Data is the only guide.