Dave's Adventures in Business Intelligence

I attended (and graduated from) Louisiana State University some years ago. During my time there they did something interesting to the quad outside of the main library. They took up all of the landscaping and planted only grass. There were no sidewalks. My initial reaction was one of disbelief, as I knew that students would be walking back and forth across the grass, wearing out paths wherever they walked. Sure enough, after only one year there were clearly defined paths in the grass. What happened next?

LSU paved those areas.

Smart.

They waited until they were able to observe established usage patterns and then they built sidewalks. In other words, rather than wasting time and money trying to anticipate how the quad would be used, and what the traffic patterns would be, they simply opened up the entire area and watched. Only after they had collected this information did they move ahead with the final stages of their project.

More often than not, I have seen companies end up taking the opposite approach to business intelligence solutions, and that hurts everybody.

When Business Objects first released their Explorer product (initially called Polestar) I was underwhelmed, to be honest. It was very expensive, and the way it seemed to be positioned originally was for the “privileged few” at the top of the food chain. That meant that the cost per seat was high and frankly I didn’t see the benefits. This was also the platform that they used to show off their natural language search engine. It was interesting, but wasn’t enough to make the cost easier to take.

Years later I saw another demonstration. During the time that passed from my initial exposure to the product memory technology had changed, prices had dropped, and the overall solution was much cheaper. That made the product slightly more interesting. The turning point for me was when one person told me that they used Explorer to define reports. When I asked for more detail, I got a very interesting response. Imagine the conversation went something like this:

Me: Wait, you use Explorer to define reports?
Him: Yes
Me: So how did you do that?
Him: When we released new data, we didn’t build any reports. Instead we simply gave the users access to this data via Explorer and let them go play with the data.
Me: So how did that help?
Him: Over time we were able to track what parts of the data were the most interesting based on usage patterns. We were also able to go back to the users and find out what views of the data they found the most useful. Many of those ended up becoming specifications for reports on the new data.

Or something like that. 🙂 The conversation was several years ago. But ultimately it presented Explorer to me in a brand new light. And of course today we have HANA, which has taken Explorer and extended it immensely. Ultimately it made me think back to the way LSU decided where to put the sidewalks. Pave where they walk. Build reports on the data they use. Makes sense, right?

Today the LSU quad still shows the sidewalk pattern they laid down years ago. The trees are older though. 🙂