In a perfect world, we'd know the source of every penny of revenue we receive.
For those of us who don't live in a perfect world have to make assumptions. One of our challenges occurs when we have to allocate "uncoded demand" to the source that theoretically drove the demand.
Nowhere is this a bigger challenge than in online orders that appear to have been generated out of thin air.
Catalogers love to create a vast array of business rules to attribute orders to one of a veritable plethora of prior catalog mailings. Catalogers frequently take too much credit for online orders.
E-mail marketers completely miss the boat here, by and large. Open rates and click through rates and conversion rates, via e-mail analytics tools and web analytics software seriously over-count or under-count orders.
Search marketers also fail to attribute orders properly. A Google query resulting in a paid search visit on a Monday that results in a purchase on a Wednesday is seldom attributed properly in the online marketing world.
This brings us to the topic of Multichannel Forensics.
Orders are categorized into theoretical "micro-channels". The phone order attributed to a catalog is a micro-channel. The online order with a catalog key code is a micro-channel. The e-mail order is a micro-channel. The paid search order is a micro-channel. The natural search order is a micro-channel. The affiliate order, shopping comparison site order, the banner advertisement order --- they're all separate micro-channels.
And yes, uncoded online orders are a separate micro-channel.
The key is to see what customers using one micro-channel do next. What is the behavior of customers who placed uncoded online orders? Do they use paid search next? Do they use a catalog key code next?
Subsequent behavior can help you understand how uncoded orders are really being sourced. If you find that subsequent orders are attributed to e-mail marketing, you have a strong sense that e-mail marketing is playing a role in the generation of uncoded orders.
Kevin, after creating a set of micro channels, would you then attempt a match back process on (any of) the orders?
ReplyDeleteThere's no reason why you can't do that, it's an appropriate approach.
ReplyDeleteI view the world a bit differently, in that I'm generally looking to see what folks do in their next order.