Attribution modeling: What it is and why marketers can't ignore it.

For years, marketers have faced the challenge of determining which type of advertising gets the "credit" for a sale or a conversion when multiple channels have contributed to pushing that person farther along in the sales funnel.

Neustar and Advertising Age released a report on an advertising industry study they conducted regarding digital attribution:
The advent of digital media promised a great leap forward: not only the ability to monitor what ads consumers saw and responded to, but potentially to make a direct connection from the advertising to the purchase itself.  
Businesses could actually count the visitors to a website, the number of searches, the clicks on an ad and the “likes” on a Facebook page. But despite the promise and the progress, and despite an overwhelming amount of information available, measuring the value of advertising on digital media channels is complicated and challenging—especially in terms of making the connection back to the actual sale.

The report acknowledges that the customer journey has become much more complex in recent years, with social media and mobile activity booming.

Enter the concept of "attribution modeling" or "digital attribution measurement." According to Paul McLenaghan, VP-product management for Neustar, it is becoming increasingly important for marketers to report on and analyze which channel drives an action or impacts the behavior of a consumer, and be able to identify and track a single consumer across those multiple channels.

While the majority of marketers (55.2% of respondents) are not familiar with the terms "cross-channel attribution" and "digital attribution measurement", according to the report, they overwhelmingly describe themselves as:

  • highly engaged in online advertising 
  • very interested in reaching the "right" online audiences 
  • wanting to spend money more effectively
  • frustrated by lack of industry consensus on the best practices
  • needing more budget and staff to make produce more accurate measurement and analysis
Sound familiar?

7 Types of Attribution Modeling

How are marketers measuring attribution and assigning credit to digital marketing channels today? The report cites 7 methods, in order of popularity, with #1 being most popularly used:
  1. Channel weighting based on the path of the customer
  2. Marketing-mix modeling / Algorithmic - a number of steps that create a formula for the marketing mix
  3. Custom weighting - with business rules
  4. First or last click
  5. Time decay - greater weight given to more recent activity
  6. Even weighting - equal credit to all channels
  7. Other
15% were unsure about the method their company employed.

"Many experts consider channel-weighted attribution an arbitrary and outdated way to assess media performance and make future media buys," says the report, so it's disappointing that this is the most popular method.

"In the 10 years since cookie tracking was adopted, the way the ad servers work is that the last exposure or last click gets the credit for the sale,” says Michael Kaushansky, exec VP-chief analytics officer for Havas Media North America.

The report also addresses the nightmare that many marketers face when trying to even track a single user to begin with. They cite the following commonly known problems with cookie-based tracking:
  • people replace laptops or computers/machines often
  • computers crash and are restored
  • computer owners know how to delete cookies and prevent cookies from being stored

What's the best solution?

According to this report: 

"For many experts, the better solution is cross-channel attribution: a method that can find unique customers across different media—ideally both online and traditional media—and ultimately connect those consumers with their real purchase data, whether those purchases happen online or in stores. 
Being able to see how consumers respond to and are influenced by the frequency and styles of advertising on various media can help marketers make future spending decisions."

The report is called "Attribution: What Marketers Need to Know About Audience Measurement."