Is Dimensional Hierarchy the Key to Web Analytics?

The process that a web analyst uses in developing a WA report and the process that a client manager uses while looking at a WA report are sort of diametrically opposite.

Someone looking at a WA report will probably always start with a KPI and then work downwards (using various dimensions for segmentation) to narrow down on specific items – for instance, if conversion (KPI) is decreasing, let’s look at various sources that are bringing us traffic; is there a specific persona that we are losing on? etc. etc.

On the other hand, a web analyst while developing a report will almost always start with dimensions and work upwards to KPIs –  five sources are bringing us traffic and our website has items to engage them, how do we measure (or define KPIs) for assessing the level of engagement and conversion etc. etc.


This raises an important question for a web analyst – where do you begin and is there a dimensional hierarchy that you can follow? In REAN model, the starting dimension is customer lifecycle; in 2-phase segmentation by Semphonic, the starting dimension is VisitorType (closely combined with VisitType); in Avinash’s model, the dimension concept comes much later (in fact, his entire model appears to have been developed from a client manager's perspective, which is great in that aspect)

Hierarchy may not be the most important thing here but the point where you begin is really important. Starting with customer lifecycle appears to be a more customer-centric approach to me. The various stages of customer lifecycle (REAN) become the base dimension to build your entire analysis. The 2-phase segmentation on the other hand does a great job of creating key mutually-exclusive-and-collectively-exhaustive list of segments but I’m not too sure if it’s as customer-centric as REAN is. Now, if I could combine the best of both, that’d be really nice!!

No comments:

Post a Comment