Based on the MarketShare White Paper Understand Multi-Channel Attribution Modeling.

Are you applying the best possible digital attribution model? What is that “best” model? To answer these questions, you need to start with a much more basic one: What are we trying to measure when we measure marketing? Put another way: What quantifiable goal is marketing here to achieve?

We would propose that the answer to that question is this: The goal of all marketing is to increase a customer’s likelihood to buy. All other marketing endeavors are a subset of that aim. What follows from that goal is the key focus of marketing attribution: measuring which marketing efforts are able to make consumers more likely to purchase.

This point may sound obvious. But many attribution models fail to address the increased likelihood to buy. Instead, they look to proxy measures, such as what behaviors and marketing engagements preceded a conversion. (“Last-click” is a often-cited example of this approach; but it’s hardly only example among many.)

The problem of focusing on identifying a path to conversion is that, no how matter descriptive that your "journey-map" might be, it still may not address the underlying question of whether the customer converted because of the ad. That’s especially important to keep in mind when you realize how many of the ads that precede a sale are end-funnel ads, such as search, targeted display, and email. So close to a conversion, it can be hard to tell which ads brought a customer “to the finish line,” and which seemed more effective than they actually were because they touch consumers who are already engaged.

What does an effective marketing attribution model look like? We have identified four components to attribution success:

  1. Know the baseline. Many customers would purchase without any ex­posure to marketing at all. Any attribution approach needs to take customers’ baseline preferences into account—otherwise, there’s no way to measure how well marketing has moved the needle.
  2. Distinguish the customer interest from the marketing effort. In a world of targeted advertising, the more likely a customer is to purchase, the more advertising they’ll be served. In that environment, it can be hard to know how much an ad should get credit for a conversion—and how much of that conversion is driven by an innate intent to buy. A solid attribution model must disentangle the two factors.
  3. Incorporate external impacts. Non-addressable factors such as the economy, price, competitive behavior, weather and more will inevitably influence consumers’ online behavior. Online attribution models need to take these other factors into account.
  4. Account for limits in the data. The right attribution accounts for the ways data is limited—from the fact that marketing itself influences the marketing data, to the fact that not all third-party data sources are equal. Good attribution approaches “know what they don’t know,” and accommodate accordingly.

Different marketers will approach these basics in different ways. The key to is to get the right attribution approach in place. To learn more about MarketShare’s approach to marketing attribution, read our attribution methodology white paper.