Information Technology Reference
In-Depth Information
you might be coming from an e-commerce situation where you might not have
enough traffic and conversion to do any decent analysis. the low traffic does not have
to be due to a small budget; you might just sell very expensive items. For the conver-
sion proxy, you choose something that in a best-case scenario should happen before the
actual sale. this can be an add to cart action, which must take place before a sale. no
one can have a 100 percent add to cart-to-sales conversion rate. So you will have a lot
more volume and data at the add to cart level than the actual sales level. this means
you will be able to use add to cart as a proxy for sales and as an optimization point.
you're looking for those highly correlated points as proxies for conversion when
optimizing. you're seeking events before the actual conversion. With a conversion
proxy like add to cart, you can be less concerned about other behaviors, such as how
people navigate and behave in general.
Figures 11.13, 11.14, and 11.15 show the ad hoc scenario i built for the three
paid search filtered steps: Basket (add to cart), checkout, and Sale, in Weeks 1, 2,
and 3, respectively. you will notice that the limited volume in actual sales is closely
related to the checkout step in the process. We might even be able to use this checkout
step as a decent proxy for our analysis.
When using a proxy, your goal is not to predict future behavior. We cannot
conclude that those visitors in checkout are more likely to convert. they might, but
it doesn't matter.
297
Figure 11.13
Paid search filtered ad hoc checkout scenario: Week 1
Figure 11.14
Paid search filtered ad hoc checkout scenario: Week 2
Figure 11.15
Paid search filtered ad hoc checkout scenario: Week 3
