View-through conversion tracking has been a much-debated (and often criticized) way of tracking online performance. Search marketers may not be as familiar, but if you’ve ever run a display campaign, you get the idea of a view-through conversion.
To put it simply, a view-through conversion happens when a user is cookied after being shown a display ad without a click, but later comes to the website and converts. Here’s a visual:
In this example, a user was shown a banner ad or video ad from a brand. Twenty days later the same user performed a Google search and arrived at the brand’s website and made a purchase. So the magic question is: Did the banner ad twenty days prior influence the behavior to search and purchase? Or was the search totally independent from the original ad? How do we know the ad was even seen in the first place? (Studies have shown that over half of displays ads are not even seen.)
Now I believe (and many studies have demonstrated) that a banner ad or video does not have to be clicked to be influential. So I would argue that when understood and used properly, view-through conversions are extremely helpful. Unfortunately, the view-through conversion has been abused. Since its inception and still today, many advertising agencies are using view-through conversions improperly and claiming credit for many purchases and online conversions that are nowhere near incremental (many of which were going to happen anyway).
“OK, enough already…what’s new in Google Analytics!?”
On June 19, Google announced a new feature that brings view-through conversion tracking to Google Analytics for advertisers using the GDN. Soon, in the multi-channel funnel reports, you will be able to see and connect activity and interactions from users who view (but do not click) on banner ads and YouTube video ads.
To see these reports you will need to do five simple things:
- Link your AdWords account to your Google Analytics account
- Update your Analytics tracking code to support Display advertising
- Sign up for the whitelist
- In your GA Admin section, click Data Sources and enable “GDN Impression Reports”
How Does it Work?
Google Analytics uses only 1st-party cookies, but banner ad impressions on the GDN are tracked by a 3rd-party cookie from Doubleclick. Contrary to what you might think, when you enable Display Advertising support in Google Analytics you are not switching from 1st-party cookies to 3rd-party cookies. Essentially the switch allows Doubleclick to communicate and transfer data from their 3rd-party cookies to the Google Analytics account.
So after enabling this support, when a visitor comes to your website and the GA code is executed, it can communicate with Doubleclick to see if the visitor has seen a display ad in the past.
Evaluating Conversion Impact of Display Interactions
In the Top Conversion Paths report you will now see an option at the top called “Interaction Type”, where you can now click to include Impressions as interactions in the table below. Once selected, you will see additional interactions that are represented by these icons:
Once these steps have been completed, you will begin to see impression interactions in MCF reports. For instance, in the example below we can see from the first row that 500 conversions took place by users who were first exposed to a display ad before clicking through on a paid search ad and converting.
This is the report you can use to see how many times somebody saw your remarketing ad, then came back to the site directly and converted. Isn’t that great to know!?
How to (Properly) Use Impression Data with Attribution Models
Now, the opportunity this creates is for marketers and analysts to continue using view-through conversions improperly. Using the recently rolled out Attribution Model Comparison Tool, you can now include Impressions as interactions. Here’s a tip: IMPRESSIONS SHOULD NEVER GET 100% ATTRIBUTION CREDIT. This wouldn’t happen unless you were using a First Interaction model, but it bears noting.
To best use Impression-level interactions in Google Analytics you should create a Custom Attribution Model. Base it off Linear or Time Decay or whatever you like…the important thing is the option called “Adjust credit for impressions.” I recommend this because I would not consider an impression as influential as a click, particularly if the impression was 20 days earlier, as in our example at the top of this page. In those cases, I like to credit the click quite a bit more, so I’ll downgrade impressions by multiplying them 0.2 times other interactions. This effectively makes an impression carry 20% as much weight as a click.
We can do even better, though, because under Advanced Option we can actually make some adjustments for time! Why would we want to do this? Well, it’s a whole lot more reasonable to believe an impression was effective if a click resulted in the next few hours. This is why cookie-windows are so important for display advertising. So I might use the 0.2 multiplier for impression interactions in general, but for impressions that precede a visit by less than 12 hours I’m willing to give more credit. To do this I’ll make the following settings in my model:
Note, I’m still not going to credit the impression as much as a click, but if the same user saw a banner ad and then visited the website within 12 hours, the chance that the impression was influential is much higher.
So now we have an attribution model which recognizes the role of banner and video impressions, provides a reasonable amount of credit for each one and rewards the impressions which are most likely to have produced activity. That’s a WIN!
A Couple of Tips & Caveats
- Each website and business is different, so the example numbers I used in this blog post may not apply. Consider the length of the sales cycle and consideration phase when making decisions about cookie windows and assigning attribution credit. For instance, car dealerships may want to increase the time window for impression credit to several days. On the other hand, 12 hours may even be too long for some businesses like restaurants.
- Impression data in Google Analytics is limited to impressions served on the GDN. This should suffice for many advertisers doing simple remarketing or display campaigns through AdWords, but if you are running a more sophisticated digital campaign using other ad networks, those impressions will not appear in Google Analytics.
So there you go! Google Analytics now has view-through tracking that automatically de-duplicates and does not require additional pixels or tracking code on your site! Cool, huh? Now…leave us some comments. Will you implement this to see the true impact of your remarketing campaigns? Are you using better methods to track display already? Leave us a comment below.