A look at attribution models
– A lot of online marketing revolves around encouraging a direct response from a perspective customer. We show them an advertisement, a tweet, or a special offer via email and then measure the success based on the click and then that conversion. This is what is referred to as last click attribution. The entire value of the customer’s conversion is placedon the marketing channel that delivered the customer. Last click attribution is easy to measure because it involves a very clear process. We understand what the user did right before they convert it.
However, this is a flawed method. We know that customers take a much larger journey.They might see an email, read a few tweets, check reviews online, and compare prices all prior to purchasing. Add offline influences and second screen interactions on mobile or tablet devices, and you can see how attributing all of the value to a single source just isn’t gonna cut it. So, what do we do from here? Truthfully, it’s an iterative process unique to each brand. So, you start with your last click wins and then you apply some weight to your other attributions such as your view-through conversions.
Now a view-through can be measured when an advertising platform places a cookie where a user has received an ad. Now, they don’t click the ad. They only saw the impression. But then they come to your site and buy your product. Well, that’ll be tracked as a view-though.They saw the ad but did not click it and they converted. But not all ad impressions are single handedly delivering sales. Again, there might be social media or search playing a role as well. But applying some weight to view-through conversions will help balance the attribution model even further.
With your paid advertising, you might adjust the view-through conversion by reducing the window of time that you’ll accept a sale as being attributed to an ad impression. Some brands might be willing to accept the idea that a 30 day window is reasonable. While others might prefer a much shorter time period. So be sure to review your cookie durations with any paid advertising that you’re conducting. As with most everything in marketing, the idea is to test, optimize, and test again. You might notice patterns emerge when you’re taking a loss on your search marketing.
That is to say, your cost per acquisition on a last click attribution is higher than anticipated.But if you stop your search marketing, you could see that your display advertising suffers significantly. This might suggest people see your banner ad, search for your brand, and then click on your paid results. If your banner advertisement was relatively inexpensive,your attribution model might allow you to share some of the costs making both campaigns pencil out in the green. You can boil attribution models down into a wholesome model or a fractional model.
So on a wholesome approach, you give all the credit to last click or first click. Whereas in a fractional model, you’ll spread the attribution out and give credit to certain channels that might only be assisting on the final decision. There are many ways to build out your attribution models. But most companies start out with last click attribution.