Remember the good old days when your weekly marketing plan consisted of an ad in the newspaper? When the phone rang, you knew how the customer had heard about your big sale on running shoes.
In today’s online world, your marketing plan has you in several channels at once—from social media to paid online ads, email and more, all driving to your website—and you might not be so sure where those customers are coming from. Because you need to stretch your limited marketing budget as far as you can, you need to know which of those channels are most directly influencing your audience and driving real conversions, not just clicks.
This is where Google Analytics comes in.
Attribution Modeling Explained
Google’s Attribution Modeling lets you look at the all possible paths you’ve laid out for your customers and weigh which one—or which combination—may have led to the conversion. As you know, consumer paths are very rarely straight. They can meander all over the Internet before the conversion.
Let’s talk about your running shoe sale. A visitor may have first heard about your company through a banner ad on the website for the 5K she’s training for. Or maybe an Adword ad popped up when she was reading an online article about arch support. Later on, our runner checks your Facebook page. She also checks the competitor’s page and Amazon.com, to compare prices. A few weeks later, she responds to your email campaign and uses your coupon to finally buy those running shoes. (She wins the 5K, by the way, thanks to those awesome shoes.)
So which channel—the banner ad, the Adword ad, the Facebook page or the email campaign—was responsible for the conversion?
The simple default model credits the last customer touch—in this case, the email campaign. But would you scrap all your Internet marketing and focus solely on email campaigns? Of course not. There are several attribution models, each with a slightly different way of looking at your marketing efforts, each with valuable insight to offer.
Attribution Model Types
Last Interaction – This model attributes 100% of the conversion to the last channel the customer used before converting. It’s useful as a baseline to compare to other models and shows where quick decision campaigns are successful. Also, it helps if your campaign goal is to convert consumers as a result of an ad, email, or post.
Last Non-Direct Click – This model gives no credit to the direct access and instead attributes 100% of the conversion to the last clicked-through channel. If you have campaigns that are heavy outside of the website, then this model lets you filter out direct visits and identify which marketing channel was viewed just before the conversion.
Last AdWords Click – As its name implies, this model attributes 100% of the conversion to the most recent AdWords ad clicked before the conversion. This makes sense when you’re trying to identify which of your campaigns actually led to the most conversions.
First Interaction – The First Interaction model attributes 100% of the value to the first channel with which the customer interacted. If you’ve created an awareness type campaign, or are trying to get your brand out there, this type of model would help you track the effectiveness of your efforts.
Linear – The Linear model looks at the entire conversion path and gives equal credit to each touch leading to the sale. Linear model is especially helpful when you’re using several channels—to first grab the consumer’s attention and then maintain contact and awareness over the sales cycle—and you want to know which channels actually contributed to the conversion.
Each touch-point the customer used is given an equal rating in this model.
Time Decay – This model gives a heavier value to the touch-points that occurred closest to the actual conversion. Touches that occurred weeks before the sale would not be counted, but those that happened on the same day would. If you’re running a time-limited campaign, for example, you probably want to credit the interactions that happen during the campaign.
Position Based – The Position Based model is a hybrid of the others and assigns value based on the position within the conversion path. For example, you might assign 40% of the credit to each of the first and last interactions, and split the remaining 20% among all the touch-points in between. This is useful when you have multiple touch-points that are important and you want to gauge their effectiveness.
Model Comparison Tool Customization – Google Analytics also provides a Model Comparison Tool that lets you create a custom model to compare the different attribution modeling. You can select up to three models to compare and create custom credit rules on how much each channel will be weighted. You could use this method to create more sophisticated modeling; for example, you can assign more credit to a channel that uses your brand name than to one that uses a keyword.
If you asked ten internet marketers which of these models is best, you’d get ten different answers. The truth is, the best model is the one that works for you. In the next blog, we’ll walk through which models work best for which type of businesses and why.