OKCupid was great for this in the early and middle days. They'd present a lot of analysis that showed some particular group of insights based on their data. T

OKCupid was great for this in the early and middle days. They'd present a lot of analysis that showed some particular group of insights based on their data. This was before 'big data' was on the CEO's radar and, even with the presentation of their analytical methods, had a feel of 'magic'. It also increased confidence that, if they could use their data to find out and disclose actionable ways in which people could increase the probability of success, they would be able to generate more probably matches than their competitors. A similar example, Wolfram's team has moved a number of similar posts from the journal and other channels to the blogs, and it's fun as well. They keep it light-hearted in tone but show off some of the "look how easy this is" with the system. What I like about the Mathematica examples is that they are reproducible by anyone with the software (OK, a non-zero barrier to entry there) or, with effort, the time to write the part of the stack employed in the solution. This is an awesome way to generate 'popular' interest while marketing your product. Coupling this with Wolfram waking up and offering the 'Home Edition' for some of their products and they're rebuilding a base that has been eroding since mma was de-emphasized in University educational programs.