A lot of "data" products stop at giving you access to raw data - a feed, a dump, an API that returns numbers. The harder and more useful problem is turning that raw data into something that fits directly into a business decision.
Location, then rules
Take weather data as an example. A raw weather feed tells you the temperature. What a logistics or agriculture business actually needs is: given this location and this threshold, should we change our plan today? That requires combining the raw data with business-specific rules before it's useful.
Where it plugs in
The goal of a well-designed data service is for the processed result to land directly in the application where your team already works, as an API or dataset - not as one more dashboard to check separately. This is the model behind Dara Intel's planned Data-as-a-Service platform, currently in development.