Where Should I Spend My Next Dollar?

Posted on Wednesday, August 26th, 2009 by Print This Post Print This Post

Categories - Analytics, Featured, SEM

When will attribution become a mainstay in online marketing? Most people I know agree that attribution (or influencers or whatever term you prefer.  I have some bias against the term “assists” since it insinuates a stronger positive correlation between media touch points which may or may not be the case ) is the holy grail of marketing. Attribution implies the ability to understand not only the direct contribution attributed to a particular keyword, banner or other channel you are tracking but also the ability to track other marketing efforts that contributed to the conversion be it a sale, lead, subscription, etc. Because this concept is like peeling an onion, this blog entry will focus on the attribution concept between keywords.  I’ll address cross-channel (Search vs. Display) and cross-media (Web vs. TV vs. Print) in future posts.

So in the world of SEM, people initially like the idea of bidding on a popular keyword like “new car.” But because it does not convert as well as say “Acura RL dealership location” they have a hard time justifying the spend against a keyword that on its own direct merits may not return a positive ROI. Now, no one I know doubts that being present on the keyword “new car” has some value; but most of us struggle to answer the question, “how much?”

How can we determine the value of the keyword “new car” and what should we do with that information?

So many factors impact a purchase decision:  friends, marketing, reviews, personal taste, etc.  Most likely, we will never understand all of the factors but we certainly should be able to understand some factors better. For the case of a general query like “new car”, we should be able to figure out who clicked on our ad, what they did when they clicked, where else they see our marketing messaging and when they actually bought, if they bought at all. Again ignoring all of the other data we would love to know (and there is a lot of it – subject of a future blog entry) , this information can be used to begin to paint a picture about the relative value each marketing message contributed to the final purchase of that Acura RL (a completely underappreciated car in my opinion).

Let’s assume I saw a total of 32 online marketing messages that were Acura RL related over the past 30 days (yes, I am only talking about online right now… you have to start somewhere, right?) through a combination of search keywords, banners, social media, etc. Some I clicked and some I did not, but all 32 influenced my decision to buy in some way. The question is to what degree did each touchpoint influence my decision, and thus how valuable was each message? There are a number of factors to consider including recency (with the theory being that the more recent the marketing message to actual purchase the more likely it had a greater influence on my decision), actual clicks versus impressions, relative interaction on the site post-click, purchase lifecycle, the actual marketing collateral, marketing relevance, etc.

An example of the recency impact is shown below:

influence report 2

Of course, all the factors need to be built into the model, not just recency.  And as we continue to dig through all of the data and test models, we will provide updates via this blog about how our thinking evolves over time.

We are already testing making concrete bid changes in paid search for a few in-house test accounts and a few of our Partners and we are excited about the results.

Stay tuned, it will be a great voyage, particularly in an Acura RL…

Tags | , ,

Leave a Reply

Follow Us on Twitter

Authors