The Primary Metric determines how each experience is measured and what defines success. It affects your experience differently depending on your traffic allocation method:
- A/B Test: A winner is declared when its primary metric reaches significant results.
- Dynamic Allocation: The variation with the highest performance in its primary metric will be served more users.
The Primary Metric can be one of the following:
Click-Through Rate: Clicks on the variation's HTML. This is useful for tracking engagement. This is supported for campaign types that serve HTML only (like Dynamic Content or Overlay).
If you use a control group, clicks are tracked for it only if the Dynamic Content or Recommendation campaigns replace an existing element. In this case, the clicks are tracked on the element that is set to be replaced.
- Purchases (e-commerce sites only): The number of purchases.
- Revenue (e-commerce sites only): The total revenue of the purchases.
- Applications (financial institution sites only): The number of applications initiated.
- Submissions (financial institution sites only): The number of applications submitted.
Goals: Any other specific user interaction that you previously created a goal to track. Goals can be based on events (Add to Cart, Submission, and so on) or based on unit clicks (to track clicks on elements that are not inside the variation). Learn more about Goals.
If the goal has a monetary value, you can choose between conversions and revenue metrics. Meaning, how many times it converted vs. how much revenue was generated from all of the selected goal conversions.
Note: In API campaigns, when the primary metric is an event (like purchase or submission), you must also decide whether the user engagement is explicit or implicit in the campaign variation (for example, a click on a special offer is necessary for a down-funnel purchase to be attributed to that offer variation). By default, it's enough that the API served the variation (the Attribution Window in the Experience OS console). Learn more about the Reporting Engagement API.
Note: In Dynamic Yield reports, you can see any secondary metric, without defining it as a primary metric, and without defining it as a secondary metric in advance.
Choosing the right metric
Revenue is not necessarily the ideal metric for every campaign, even for e-commerce sites. We recommend choosing a primary metric connected to your experience hypothesis and the location of the experience in the funnel. For example:
- CTR for a hero banner on the homepage
- Add to Cart for a social proof notification on a product page (e-commerce)
- Number of applications for an email capture overlay (financial institutions)
Choosing a metric that reflects the action you want from the user at this stage of the funnel will lead to faster conclusions because:
- We will have fewer variables in the data. If a test is on the homepage, and you use a primary metric that occurs later in the funnel (like submissions or purchases), there are more variables involved because the user's journey involves additional steps after visiting the homepage. This means that the algorithm takes longer to determine which variation is better.
We will have more data points.
E-commerce example: If you're testing a new product page layout, the experience should measure how often users react by adding something to the cart. If you set the primary metric to revenue, you'll miss out on data about users who added something to the cart but didn't make a purchase.
Financial institution example: If you're testing a new hero banner on the homepage, the experience should measure how often users react by starting an application. If you set the primary metric to submissions, you'll miss out on data about users who started an application and didn’t make a submission.
Having fewer data points results in the algorithm taking longer to conclude which variation is better.
Improving your conversion funnel should be done one step at a time. While keeping in mind the overall goal of your funnel, you should design and measure experiences depending on where they are in your funnel, and what the goal of that particular step is.