Thursday, May 29, 2008

Performance analytics the Coradiant way

What has now been recently rechristened "performance analytics" has been in existence for decades in IT: managing availability, capacity and performance. I've been a system administrator, dba, developer of real-time applications for the Montreal stock exchange and involved with the Web since the first incarnation of Mosaic, so I was comfortable meeting Coradiant at their R&D center in Montreal to discuss about the new TrueSight WA integration with Omniture SiteCatalyst (through the Genesis plug & play architecture).

A case for multiplicity

By now it's a given that web analytics is not only about visits and page views. The concept of "multiplicity" put forth by Avinash Kaushik is not yet tapped by most organizations. Put simply, here's how Coradiant TrueSight WA can help:
"captured user information is typically used by the IT department for troubleshooting, service level reporting and change management. But with TrueSight WA, web traffic, performance, and availability metrics gathered from each user visit are seamlessly integrated into Omniture SiteCatalyst, so marketers can get entirely new insights into campaign and customer success."
Coradiant integrates to your infrastructure to collect performance information directly, without additional impact on the client side (no JavaScript tags to change).

Correlation IS causation

Especially in the field of web analytics, how many times have we heard "correlation does not not imply causation"? It's often a beginner's mistake to take two distinct metrics and come up with a misled explanation of why one would impact the other. When we're talking about conversion rates and performance, there IS a very strong correlation between the two. Put it anyway you want, I can guarantee you that non-availability will lead directly to zero conversion. I can also guarantee you that poor performance will lead to lower conversion.

So why do most web analytics solutions keep offering only non-performance related statistics? I guess it's in part caused by the great divide between IT and marketing. Now, thanks to Coradiant, that gap will be a little less cumbersome, bringing more context and a similar language around IT and marketing (at least when it comes to web analytics!).

An example

Let's make the case for performance analytics with what a typical empowered web analyst would say about a campaign (fictitious case):
"Our marketing campaign increased traffic to the site by a factor of 4 times, bringing 120,000 visits to the site on a single day instead of the usual 20,000 or so. Although the conversion rate was lower (2% instead of 4%), the campaign was a resounding success, bringing $X in increased revenues. In our next campaign we will try to increase conversion rate by doing better ad placement, but still, bringing more traffic also increased our brand awareness and engagement toward the site."
With Coradiant, here's the additional insight gained:
"The campaign was a success from a marketing perspective, but since 10% of the visits resulted in errors and about 15% were slower than usual, at the 4% conversion rate, we left $Y on the table and we probably have frustrated a fair number of our visitors. Next time I would recommend spreading out the campaign or increasing the infrastructure capacity beforehand."
Unless you were very, very, very friendly with IT, you would have never known that. Even then, it would be a lot more difficult and time consuming to come up with such a clear conclusion.
Post a Comment