July 07, 2008

Google, Yahoo!, MSN, and Multichannel Forensics

Web Analytics practitioners do a very nice job of summarizing visit-specific customer behavior. They can tell you that 29% of the customers who visited a landing page were new, 71% existing visitors. Our world of marketing and measurement fundamentally changed when we were given access to the information provided by Web Analytics.

Now it is time to complement the veritable plethora of information provided by Web Analytics with Multichannel Forensics.

Most online marketers generate considerable sales from customers referred by Google, Yahoo!, and MSN searches.

Now let's consider future behavior. In other words, does the Google purchaser purchase again after conducting a search using Google? Does she purchase again because she becomes loyal to your website, visiting it frequently? Does she purchase again because of your catalog marketing activities? Does the MSN purchaser stay loyal to the MSN micro-channel, or does she switch to Google?

Multichannel Forensics provide answers to all of those questions.

Often, Multichannel Forensics suggest that Google is a unique micro-channel, separate from your e-commerce channel. My projects routinely indicate that Google customers have comparable future value, compared with Yahoo! or MSN customers. But subsequent behavior is fundamentally different! This means downstream marketing activities can be executed differently for the Google shopper than the MSN or Yahoo! shopper.

The Yahoo! and MSN shopper are occasionally less loyal to their portal than are Google shoppers. Loyalty to your brand may be similar, but you may find these customers elsewhere on the internet in the future, whereas the Google shopper may continue to be loyal to Google.

Results across brands are frequently unique, so it is important that you give Multichannel Forensics a try, considering each online marketing vehicle as a unique micro-channel.

The Math Behind Best Customers

Those of you involved in Catalog Marketing run optimization tables, right? Right?? Here's an optimization table for really go...