March 06, 2009

Non-Linear Direct Marketing

Last week, I made a comment about online marketers ... "they're the folks who stare at you when you mention anything that cannot be immediately quantified with a conversion rate metric". Ted asked for clarification, suggesting that conversions, from a direct marketing standpoint, were a good thing.

It is my belief that direct marketing has always been non-linear. If direct marketing were linear, cannibalization would not exist.

It is my belief that direct marketing has always been treated as a linear process.

For instance, we've always talked about the purchase funnel, with linear concepts like "aware-consider-shop-buy-own". Given this linear process, we set up our KPIs, our "metrics", around this purchase funnel. Does the customer visit more than just the homepage? Does the customer put an item in the shopping cart? Does the customer buy something (conversion)?

These metrics allow us to measure success. The online marketer is uniquely qualified to measure each step in this process.

But ask the online marketer to answer any question that is non-linear, and the process falls apart. For instance, ask your web analytics expert the following set of questions:
  • Of all the customers who abandoned a shopping cart last week, what percentage visited the site again this week? And of those who visited the site again this week, what percentage purchased something?
You can answer this question with many web analytics software packages. That being said, few web analytics experts can answer this question for you. Even fewer online marketing experts can answer this question for you. If you ask this set of questions, you are very likely to get a blank stare. Go ahead and try it sometime!

This is important, because a customer originally labeled as a "failure" by a linear set of metrics (the customer abandoned a shopping cart) becomes a "success" because of a purchase the following week.
  • Our Metrics (Linear): Shopping Cart Abandonment = 50%, Conversion Rate = 50%.
  • Actual Outcome: Conversion Rate = 100%.
Now granted, this process is technically "linear" according to the definition of a purchase funnel. But the behavior of this individual customer does not fit our linear set of KPIs --- a successful outcome is actually recorded as a series of unsuccessful steps.

This is one reason that I typically review annual timeframes with a Multichannel Forensics analysis. We can see interactions better, we can see if a series of negatives result in a positive, we can see if combinations of advertising fundamentally change customer behavior.