Read between the lines – this expression derives from a simple form of cryptography, in which a hidden meaning was conveyed by secreting it between lines of text. Unfortunately, web analytics tools often limit your possibilities to extract meaningful data by combining data from separate rows. Let me exemplify:
Internal bounce rate
You don’t want your customers to end up in dead ends on your web site, right? One sign of dead ends is what we can call internal bounce. Imagine that a visitor on your homepage notice a push for certain product, let’s say Converse sneakers on sale, and clicks on the push. However, when the visitor enter the landing page he sees that the sale is limited to kids converse sneakers and hence is not relevant for the visitor. Thus visitor chose to go back to the homepage. This is an internal bounce. The interesting information here doesn’t exist on a single logged hit or row, but three. It’s only when we study those together we can identify that the push doesn’t communicate the same message as the landing page. So, how do you get hold of an internal bounce? Well, if you define an internal bounce as page view where previous page is the same as the next page, then it’s only a matter of adding previous and next page as dimensions to your page view level. That can be done by window functions or self joins. But it is hard to accomplish in traditional web analytics tools, if you collect a lot of data you probably have to export web data to a MPP or run a big data solution such as Hadoop or Google Big Query.
Product conversion rate (calculated metrics)
Often conversion rates are calculated as a ratio of “sessions with transaction divided by sessions”. Usually it is possible to segment by campaigns and if you are lucky you even have support for attribution modelling. However, this kind of conversion rate doesn’t help you optimize yield since it is completely disconnected from products. You only know the quantity sold of a particular product, but not the figures behind. Let’s say you don’t reach sales target on a product A, what should you do? Lower the price, promote on homepage and e-mail, change product copy and images or check product availability and errors? Sales figures doesn’t tell you, but product exposure and product conversion rate give you a hint. If exposure is low for product A but conversion rate is high, then it may be a matter of promoting the product rather than lowering price that is the solution. The opposite is also true, high exposure and low conversion rate imply you may have to lower price or work on your product presentation. Otherwise you will give away valuable profit margin unnecessarily. The exposure and the transaction are logged as two different events and web analytics tools doesn’t allow you to create the calculated metrics needed to tie those together.