October 30, 2007

How Online Advertising Channels Fit Together

Please click on the image to enlarge it.

E-commerce provides us with the richest array of information ever available to direct marketers. Let's use the Multichannel Forensics framework to explore the interaction between advertising and customer purchases on your website.

Here's what you do.

Step 1: Create a "spreadsheet", if you will, with each row representing a customer who purchased in 2005. Each column in your spreadsheet is populated with a series of "1s" or "0s". Each column represents a form of advertising. A one is populated if the customer purchased in 2005 using that form of advertising. A zero is populated if the customer did not purchase in 2005 via that form of advertising.

A list of potential "advertising" channels include:
  • Brand Buyers: Customers who purchased, and we cannot identify any advertising campaign that caused the customer to purchase. This is the most important "channel" ... we want our customers to purchase from us because they "love" us, not because we advertise to them.
  • Catalog: Customers who use a catalog source code, or are identified as having purchased via catalog from a matchback analysis.
  • E-Mail: Customers who clicked through an e-mail and purchased merchandise.
  • Social Media: Customers who purchased and had a referring URL from a blog or social media site like MySpace or Facebook. Some companies comprise a list of the top 5,000 referring URLs, then visit each URL, and identify if the URL is a blog/social-media site. In addition, these companies look for URLs with phrases like "Blogspot" or "Typepad" in the URL, to identify that the referring URL is a blog.
  • Portals: Customers who purchased and had a referring URL from an advertisement on a portal like MSN or Yahoo!.
  • Paid Search: Customers who purchased and had a referring URL from a paid search ad.
  • Natural Search: Customers who purchased and had a referring URL from a search engine, a URL that isn't from a paid search campaign.
  • Affiliates: Customers who purchased via an established affiliate.
  • Shopping Comparison: Customers who purchased via a shopping comparison site like MySimon.
Step 2: Repeat Step 1, but this time, create the file for calendar year 2006.

Step 3: Match the files together for Step 1 and Step 2.

Step 4: Conduct a Multichannel Forensics analysis.

Step 5: Create the map at the start of this post. Any channels that are in "Equilibrium" or "Transfer" are linked together on the map.

The analysis illustrates how your advertising channels work together.

In this example, there are numerous findings.
  • New online customers are recruited through paid search, catalog marketing, and "brand buyers", customers who purchase without having used advertising.
  • Shopping Comparison and Affiliate customers appear to be largely "cut off" from the ecosystem. In other words, these customers do not eventually become customers who purchase "on their own", or via other advertising channels. One would evaluate the LTV of these customers, to understand if these marketing activities are worthwhile.
  • Social Media sites and Portals are leveraged by "brand buyers". In other words, best customers utilize these sites to make purchases.
  • Paid Search and Natural Search become "links" that ultimately facilitate the transition from new customer to a loyal customer. The data in this example suggest that these customers can eventually become "brand buyers", customers who don't need advertising to purchase.
  • Catalog Marketing is an important way to acquire new customers. The data also suggest that catalog customers use search to complement future purchase activities.
  • E-Mail is a "dependent channel" in this example. E-Mail marketing might be effective, but it appears it depends upon other marketing activities (search, catalog) to "acquire" the e-mail address that will be used for e-mail marketing purposes. In this example, the link between "brand buyers" and e-mail is one way. Brand buyers offer an e-mail address, and then respond to e-mail. E-Mail customers are less likely to become "brand buyers", in this example.
Now that the advertising ecosystem has been fully mapped, you sit down with your marketing folks, and consider appropriate linkages between marketing programs.

It is also important to understand which advertising channels drive a customer toward loyalty. Notice that catalog, search and e-mail marketing are all interconnected, whereas affiliate and shopping comparison marketing activities deliver a fundamentally "different" customer to the brand (in this example ... your mileage may vary).


  1. Anonymous2:09 PM

    It has been my experience that online affiliates can grouped into major types with different behavior and customers associated with each. I’m sure your readers have developed their own taxonomy, but here are a few more nodes for consideration within your ecosystem:

    • Shopping Engines and Promo/Coupon Sites (shopzilla) – Primarily product and price driven buyers with little brand affinity, low repurchase rates, and low subsequent value

    • Content and Community Sites (photodo) – Primarily product driven buyers with some brand affinity and decent subsequent value.

    • Loyalty and Reward Sites (Upromise) – Primarily brand and offer driven buyers with high repurchase rates and strong subsequent value.

    • Cause-Related and Charity Sites (nwf) - Primarily brand and offer driven buyers with high repurchase rates and strong subsequent value.

    Of course, each business and offer is different in terms of the interaction of the ecosystem’s channels, but you may find that some affiliate marketing activities might not deliver a fundamentally "different" customer to the brand.

    Casey Carey/Epsilon

  2. Well done, Casey!

    Have you used your expansive mountain of data to conduct a Multichannel Forensics analysis?

  3. This is a great post and i have definitely learned a few new things about online avertising and creating campaigns...it almost makes me a little nervous to attack it this way!

    Thanks again!


Note: Only a member of this blog may post a comment.

An Example of Hopping on to Chat With Your Community

Do you have a community? Do you have one off-platform? Here's an example of a Sennheiser product manager hopping on Reddit to chat with ...