STATISTICA Blogs

A friend recommended that I read Avinash Kaushik's 2009 book called Web Analytics 2.0: The Art of Online Accountability and Science of Customer Centricity.

I love the sense of humor and practical approach that the writer displays. There were sections that reminded me that Web Analytics has many similarities to "classic" Statistical Analysis. And some sections that made me think "data mining could solve that problem".  

For example, the author talks about the paradox of data:

"a lack of it [data] means you cannot make complete decisions, but even with a lot of data, you still get an infinitesimally small number of insights"

My knee jerk reaction to this was... data mining was made to solve the "too much data" problem. If you can link your website data with CRM data, then you could find even more insight.

Another example, the author wrote:

"Perfection is.. the Enemy of Good Enough

Data quality on the Web is not perfect; things change too fast, everyone wants a piece of data yesterday, and your competitors are strong... If you have 90 percent confidence in the data, then make a decision. Don't wait for perfection"

I think this applies to everyone. Anyone have perfect data? Data cleaning (dealing with missing data, duplicates, etc..) is a very common first step in data analysis. 

I can't find the quote right now, but there was another section that resonated with me. The author said that if your reports/data analysis aren't acted on, then it is time to reboot. 

This is a great point... Reports should be useful and used. If they aren't, then it is time to explore other options.