A Look at Clario Analytics as a Service

As part of a larger BI package exploration I took a look at Clario Analytics SaaS (Software as a service). You start by uploading your data via a secure FTP client to Clario. That process was easy, but I was not fully assured that my data had arrived at the right place. The interface is then designed to help you create a data flow diagram, which I have used before. The user interface is a somewhat nonstandard, minimalist and hard to figure out without some initial help. I was assured that they plan to change it radically to a more standard interface in a few months. Good idea.

I then also attended their introductory Webinar. They make some good points about Saas in general. It keeps you from having to maintain software and worry about how much space you need. No installing software. Backups done automatically. It also lets you set up a library of models you are using. All good, though I also want to make sure that the company and methods are stable if I entrust my analyses and data to them.

When I first looked at the underlying analytics it looked simplistic, but the webinar showed me that there were a number of advanced methods, including nodes that select the best analytics for deriving relationships, via branch-and-bound methods. A little concerned that the average user could not carefully determine statistical validity for their problem. I have not looked at the underlying methods yet, they are in their wiki-style documentation, which was easy to navigate.

You run the flow diagram in a batch mode, it runs on their processors. My simple 10MB data analysis took maybe 30 seconds. Here is another advantage of Saas, they upgrade the processors. You then open up the results. I really like to see the output visualizations immediately and would have preferred to see them appear without me asking. I Would also like to see any diagnostics as well to make sure I don't misinterpret results. Flagging analysis problems would also be useful, did not see that. They have a number of analytical consultants, so you can outsource the design of your predictive analytics as well.

Some useful case-studies with impressive ROI. Most of the examples they give are in the direct mail domain, but this seems it would work well for many relatively simple analytics applications. You pay as you go per month or yearly. I imagine many small to medium size enterprises have intermittent need for Saas analytics, and this would be a good place get that. Also a good example of how Saas systems work. Clario is worth taking a look at.