Sending an expensive B2B sales team out to discover what we should build isn’t a great strategy. We should do less expensive, unemotional, non-commissioned validation and learning before scaling up our selling effort.
There are no generic or universal KPIs, since every business has unique aspects. So if we want KPIs for a B2B/enterprise company, where would we start? And how do we avoid committing to improvements in metrics/KPIs before understanding our current scores (or situation)?
As product folks, we should be responsible for reasonably anticipating misuses of our products, as well as harm that flows from fundamental product/economic goals. It’s not clear how we step up to this, though.
Product managers working with data science teams on production applications have more challenges than with more deterministic (traditional) applications. These include providing more business/user context, not assuming that data will be predictive, and discussing accuracy requirements at the very start of a project.