Adobe Target Activity Impressions Metric Explained
“What is the clickthrough rate (CTR) for this offer?” – Anyone who has used Adobe Target (AT) to serve marketing offers has probably been asked this question. It’s also likely that the answer hasn’t been straightforward. So why is such a basic metric so problematic for Target? The biggest problem is the denominator: Impressions.
First, let’s simplify and assume a brand-new basic implementation of Adobe Target without any custom analytics tracking to go with it. The Target library (at.js) is on the page, edits can be made in the Visual Experience Composer (VEC), and HTML offers can be created and swapped in and out at will. Enter two marketing managers, Bill and Ted.
Bill creates a new high priority Activity in the VEC with a single Experience which replaces the homepage hero banner for all existing customers. He sets the banner CTA click as the goal, tests it out, activates, and watches the conversions roll in. With a couple clicks in Target, Bill can see impressions, conversions (clicks), and a conversion rate. Simple enough.
Ted also has a lower priority Activity he wants to run in the same homepage hero space targeting only existing customers in California. He sets up his Activity in the VEC with the proper audience and activates it. After a few days Ted notices that his Activity is receiving impressions but no conversions. Bogus… that can’t be right. What is going on?
For better or worse, this behavior is by design.
The impressions tracked by Adobe Target are the Activity Impressions, which really mean how many times AT returned a response for that particular Activity and Experience. An Activity Impression does NOT mean that the user actually saw the content. In the case of Bill and Ted, both of their homepage hero offers occupied the same space but were targeted to different audiences. Ted’s audience was a subset of Bill’s and as a result only Bill’s offer was actually shown to visitors. It makes sense that only one offer can display for a particular location, but why were impressions counted for Ted’s Activity at all?
Using the “global mbox” Target will return all Activities and Experiences that the user qualifies for and will NOT deduplicate if they are modifying the same CSS selector. Unless you have a customized Target implementation, then any modifications made in the VEC are delivered by the global mbox. Because Bill and Ted are both targeting the homepage hero, both of their offers are technically returned from Target and are receiving impressions.
Why does Target count impressions this way?
Let’s take one more step back and look at Target’s history. Target has its roots with displaying offers to specific audiences in predefined locations known as regional mboxes. Adobe has since evolved the product to enable manipulation of pretty much anything on the page and has slowly let the concept of regional mboxes fall by the wayside in favor of flashier capabilities. Basically, Target is an experience testing and personalization platform. The keyword is experience. Target’s workflows and reporting are designed with the experience at the forefront, not individual offers. The flexibility to control experiences that span across multiple pages and locations requires out-of-the-box metrics that also support that paradigm. So, as a result, that’s why we have “Activity Impressions” instead of “Offer Impressions”.
With Target, an offer can be anything. It can be a chunk of HTML for a carousel, a reference to a static image, a change to a CTA URL, a rearrange of page components, a custom script injected into the head of the page, etc. Now that the definition of an offer is broadened, it makes sense why the definition of an impression must broaden with it. While I understand why impressions are counted this way, for new (or even veteran) users of Target the technical definition of an Activity Impression can be easily overlooked. If you aren’t careful, there can be serious analysis missteps when working with Target’s Activity Impressions metric, or any other calculated metric based on it.
Although macro-level metrics for a test or experience are important, there is still a place for more granular tracking of offers. In future blog posts I’ll cover activity overlap in more detail, different tracking approaches to avoid the impression overlap issue, and explain why “regional mboxes” can still be very useful.
