Thanks. What startups can do is label and organize their data in advance of bringing in a consultant. That doesn’t require a PhD. Usually they have one or two main tables to learn from, and the better they tee it up for me, the less the project is likely to cost in terms of slippage and risk. In my view data quality and quantity are inversely proportional to risk.
In big data, we get volume variety and velocity. In a startup with volume but not variety and velocity, solutions can be developed and deployed quickly.