[trussi]

Take a look at Tableau. It's super expensive and desktop-based. But it allows for some very powerful, customized data analysis.

Build a SaaS version of Tableau and you have a good starting point.

You still need to find the problem you are solving. Try to find one or two niches, like university labs or government research facilities.

Personally, I'd find the problem this solves well, then actually solve that problem (instead of helping others solve that problem). Take credit card interest rates; instead of building a tool tailored to credit card issuers or resellers, build a site that provides the data (using your tool) in a value-added way.

It's like you're creating your itch, so you can scratch it with your idea. A bit backwards, but crazy enough it might just work.