Five Options for Attribution Modeling With Analytics Engines
Originally published on MarketingProfs on December 11, 2014
Listen to the talking heads covering football broadcasts and you’ll often hear them mention “half-time adjustments.” What they mean is one team’s ability to learn its opponent’s tendencies and preferences, then change what it’s doing to counter their opponent. The winning team often isn’t the one that walked into the stadium with the best game plan but the one that saw what was happening on all points of the field and changed its game plan accordingly.
Marketers have their own version of halftime adjustments, only we call it “attribution modeling.” The objective is to evaluate (by using analytics) what led to every sale—so we can replicate success.
Traditionally, marketers have relied on one specific model, “last click,” because that was all that was readily available to us. Today, however, thanks to free tools in modern analytics engines, marketers have many more available options.
Unfortunately, that doesn’t necessarily mean we use them. In fact, research by Google suggests that although more than 90% of marketers have advanced attribution analytics available to them, only slightly more than half are taking advantage of them.
Part of the reason may be that organizations aren’t aware that the tools exist. Others may be concerned that the tools are too complex. Another reason may be that some have tried previously but failed.
But the tools are worth learning because they are like a superstar quarterback on a football team—a competitive advantage that is difficult to overcome.
Though many different methods for attribution modeling exist, the following five command the most attention or offer the most promise.
1. Last click
This is the de facto standard in marketing, most likely because it’s the easiest to track. This model shows the last action the prospect took before making a purchase or becoming a sales lead (depending on the marketer’s objectives and sales cycle). Did the prospect click on paid search? Organic search? A banner ad? No effort is given to discovering how the prospect got to that point before the desired action—often because marketers aren’t sure how to track the full trail.
Using last click is equivalent to trying to put together a football game plan by watching only the scoring plays. You’ll get an idea of what works in the red zone… but not much else. Focusing your marketing efforts based on last click might be the right move—or it could be a tremendous waste of budget. There’s no way to be sure. Last click may have been the only game in town at one time, but marketers now have many better options.
2. First interaction
In this model, you’re placing all your focus on discovering what caused prospects to become interested in your products or services in the first place. Did they click on a banner ad? Search on specific keywords? However they got there, this view is just as limiting as last click. You still have only one touchpoint on which to base your marketing decisions—it just moved to an earlier point in the process. Now, instead of watching scoring plays, you’re watching kickoffs and punts.
Now we’re getting somewhere. With this model, you’re accounting for all the different positions in the sales funnel and weighing them based on what you believe to be most important. For example, you may give 40% of the weight to first interaction, 40% to last click, and 20% to what happens in the middle. The key is it’s showing every single touchpoint throughout the sales cycle.
As you learn more about what motivates your customers to action, you can adjust weighting to better reflect how they interact with your brand. You can set up multiple customer models to give weight to different touchpoints, too. You also can perform a great deal of customization based on what’s important to the sale. For example, you can make adjustments for time on your site, pageviews (the more pages viewed, the more credit a particular channel gets), position rules that weight the value of a conversion based on each interaction, and more.
You’re seeing the whole field and the whole game—beginning to end.
As the name implies, while you’re still looking at the entire sales funnel, you give more weight to actions that occur closer to the conversion. So, for example, while some credit is given to an organic search that occurred 30 days ago, it is not nearly as much credit as the banner ad that was clicked two days ago, or the paid social media ad that was clicked today. Consider it your two-minute drill.
This model is incredibly powerful, but also requires the greatest degree of expertise. Rather than working from a starting point, you are customizing everything, with many more parameters available to you. Really, it’s the best of all the other models, combining elements of positioned-based and time-delay.
For example, you can look at media in the past and the actual media type. If you feel paid search is more important than organic search, you can adjust the weighting that way. You can look at the path to get to a conversion and/or you can look at the different interactions. The problem is that, with all this capability at your fingertips, it’s easy to get lost and confused. Your best bet is to start with one of the other models, and as your competency grows move into linear attribution marketing.
What does all of this capability do for you? In the end, it tells you what’s most important to your sales cycle, comparing the rest of the path to the last click, so you know how to spend your budget. For instance, last click may tell you that a particular keyword is worth $5.46 per click based on sales, but when you look at a position-based model, you discover it’s really only worth $4.98. Having that information will prevent you from bidding up that keyword, allowing you to use your budget to influence other positions important to the sale.
* * *
All you need to do is invest a little time into learning how to use these new models. And then continue to tweak, tweak, tweak as you learn.
Having a great game plan is important. But half-time adjustments often determine the winners. Learn how to use the various attribution modeling tools available. and you’ll be in a position to claim more victories.