Guest Blogger Eric Krell: What Is Attribution Modeling?

Customer Engagement
Customer Experience
With pressure mounting on the marketing function to more clearly identify the return on various investments, attribution modeling may seem like the Holy Grail to some marketers.

With pressure mounting on the marketing function to more clearly identify the return on various investments, attribution modeling may seem like the Holy Grail to some marketers. To others, attribution modeling--or attribution management, as it's often called--is as much an enigma as the concept of marketing ROI.As someone who has for years written about ROI for an audience of CFOs, I am intensely curious about attribution management. I researched and wrote "Give Credit Where Credit Is Due" for 1to1 Magazine to better understand how attribution management uncovers the influence each marketing touchpoint has on an online sale.

In the hopes of both stimulating and quenching your curiosity, here's is an excerpt of a conversation I had with C3 Metrics CEO Mark Hughes, who couches complex attribution concepts in more accessible terms:

What is attribution modeling?
It's best described in context of the problem it solves.

Imagine a simple Internet purchase: a $100 transaction from A reader on The New York Times website sees a display banner for Zappos. It makes her think about sandals. She surfs on Zappos and other sites. After a beach weekend, she decides to order the sandals she looked at by typing "Zappos" into Google, and makes a purchase. One hundred percent of this advertising credit goes to the paid brand search ad, which was simply the endpoint before checkout, not the starting point: the ad that first stimulated her purchase. The marketing executive running the Zappos' Internet advertising budget gets a false read, with misleading data on which advertising expenditure truly drove revenue.

Today, Internet ad tracking systems are predominantly the same legacy systems from the late 1990s. The problem with these systems: They erroneously give 100 percent credit for a transaction to the last clicked Internet ad, or the last viewed Internet ad. No credit is given to ads that create or cause a purchase and consequently no credit is given to actual revenue drivers. This "last click" attribution misleads advertisers as to where to allocate Internet ad dollars.

"Last click" is the problem that attribution modeling solves. In short, attribution modeling is the process by which fractional credit is assigned to ads bought by online advertisers at a transaction level.

What marketing activities can it be applied to?
Display, search, affiliate, email, comparison shopping engines, unique URLs or domain names--and stay tuned for...TV attribution. When a TV spot airs and a website sees online activity, [it will soon be possible to] appropriately attribute credit to the TV spot that drove people online.

What marketing activities may not be a good match for attribution modeling? Why?
Since the online industry is still measuring everything with last-click attribution--except for a handful of thought leaders deploying attribution modeling, [aka] full-funnel attribution--the natural place to start attribution modeling is with online marketing. Anyone spending more than $50K per month online will be using some form of an attribution model by 2015.

Radio and Outdoor probably don't lend themselves to attribution modeling.

Are there pitfalls to avoid?
Yes indeed, and the biggest pitfall relates to timing. Once you start using an attribution model, which has no event limitations or time limitations, the biggest pitfall is making media decisions before an adequate amount of time has passed.

Once a marketer or agency deploys an attribution model, they'll see things they've never seen before. One of those things is median days to conversion by channel. Case in point, one company discovered that the median days to purchase for generic search terms (non-brand search terms) is 13.7 days. But they were optimizing search every seven days. Essentially they were making critical media decisions before all the data was in.

It's similar to if a quarterback throws the ball at a certain spot six steps before the receiver gets there--he's acting too early. You can't make decisions before the data is all in, and different media channels gestate differently by means of time.

Essentially, you have to think more like a winemaker, where time is a factor.

+ + + + + +

About the Author: Eric Krell is a freelance journalist and frequent contributor to 1to1 Magazine.