HIdden Variables: A Thorn In The Side Of Multichannel Customer Insight

Here's one for you.

On a recent Multichannel Forensics project, I had information suggesting that one channel performed much worse than the other half-dozen channels managed by the brand. I was ready to recommend that this company discontinue marketing through this channel.

Then I looked at another series of attributes. I discovered a "hidden" variable that caused the channel in question to appear bad. After accounting for this hidden variable, the channel actually performed above-average.

Let's look at a simple example, with just five customers.

	Future
	Value	Channels
Customer 1	$200.00	One
Customer 2	$50.00	One
Customer 3	$500.00	Two
Customer 4	$100.00	Two
Customer 5	$25.00	One
Multichannel	$300.00
Single-Channel	$91.67

This table illustrates the classic argument that "Multichannel Customers Are The Best Customers". Given the data available to the marketing department, the argument seems to make sense, right?

Now, let's overlay a "hidden" variable, one that tells us if the customer lives in an Urban zip code or a Rural zip code. Take a peek at the findings:

	Future
	Value	Channels	Zip_Code
Customer 1	$200.00	One	- Urban
Customer 2	$50.00	One	- Rural
Customer 3	$500.00	Two	- Urban
Customer 4	$100.00	Two	- Urban
Customer 5	$25.00	One	- Rural
Multichannel	$300.00
Single-Channel	$91.67
Urban	$266.67
Rural	$37.50

Oh oh.

In this very simple example, being "multichannel" isn't the real issue. The real issue is that urban customers have access to stores, and can purchase merchandise online. The most valuable customers are urban customers, because they have access to multiple channels, while the rural customer truly has limited access to stores.

This doesn't mean you shouldn't invest in a truly wonderful multichannel experience for your customers.

Rather, our "world-view" is limited by what we can see. In multichannel marketing, we can't see much!

Hidden variables require two parties to collaborate appropriately. Leadership must be able to ask appropriate questions that cause hidden variables to expose themselves. Business Intelligence teams must be able to construct queries that yield variables that expose hidden variables.

Multichannel Business Intelligence isn't the only place where hidden variables play a role. Web Analytics and E-Mail Marketing professions are marred by hidden variables. Unless gifted, highly experienced professional have an intuition for "what to look for", it is possible that the real reasons for a successful conversion rate will not be identified.

Knowledge of hidden variables is a good thing. We don't have to know what the hidden variables are. We just need to know that something is lurking behind the scenes, influencing results. We need to question what we see.