Create a metric in Optimizely Web Experimentation

  • Updated

This topic describes how to:

  • Understand the different components of a metric in Optimizely Web Experimentation
  • Create a metric in Optimizely Web Experimentation's metric builder
  • Add a metric to your Optimizely Web Experimentation experiments

See Create a metric in Optimizely Feature Experimentation for how to make a metric in Optimizely Feature Experimentation instead of Optimizely Web Experimentation? 

Optimizely Web Experimentation's natural-language metrics builder lets you quickly define and add metrics to your experiments and campaigns. It works by asking you to define a small set of parameters that tell Optimizely Web Experimentation how to measure and report the results of your experiment or campaign. These parameters include the winning direction (increase or decrease), what you want to measure (the numerator), and the rate at which you want to measure it (the denominator).

The process of building a metric using the metrics builder is straightforward.

add-exit-rate-metric.gif

  1. From the Experiments tab, open the experiment with which you want to work.

  2. Click Metrics to open the Metrics window.

  3. Choose an event from the list to be your metric.

  4. Choose a name for your metric and set its winning direction, numerator and denominator.

Optimizely Web Experimentation offers several pre-built metrics, as well as a flexible numerator called total value. If you are not sure which metric to choose, see Types of metrics and when to use them.

To learn how to edit your metrics once you build them, see Edit a metric in Optimizely Web Experimentation.

The metrics builder

The metrics builder is the interface within Optimizely Web Experimentation that enables you to create your own metrics.

metrics-builder.png

You can only build metrics in Optimizely Web Experimentation if your role is authorized to create and edit campaigns and experiments—that means administrator, editor, or project owner. See this article for more details.

Several metric types have additional requirements you need to be aware of prior to using them:

Winning direction

When you set a metric's winning direction parameter, you are telling Optimizely Web Experimentation what kind of change you hope to see: an increase in your metric, or a decrease. In most experiments and campaigns, you will want to see your metrics increase. However, for some metrics—like bounce rate, Cancel button conversions or cart abandonment—a lower value (that is, negative lift) would be more desirable.

Numerators

The metrics builder offers six different types of metrics templates. These types are also called numerators. You should select your metric's numerator based on the specific questions you want your experiment to answer: 

  • Unique conversions – Number of visitors with at least one conversion

  • Total conversions – Total number of conversions

  • Bounce Rate – Total number of times where the page being viewed is the first and only page the visitor sees before leaving your site

  • Exit Rate – Total number of times where the page being viewed is the last page the visitor sees before leaving your site

  • Total revenue – Total revenue generated (you will have to set up revenue tracking before using the total revenue metric)

  • Total value – Total of any other numerical value (check out our article on total value use cases)

For more details on each metric type, see our article on when to use each type of metric in Optimizely Web Experimentation.

If you are using total revenue or total value as your numerator, complete an additional step. The additional step for total revenue is not the same as the additional step for total value. Make sure you are reading the correct section.

Denominators

The metrics builder's denominators can also be thought of as the rate at which your metric measures its numerator. For example, you could measure unique conversions per session, or total revenue per conversion.

Because of the inherent differences between metric types, and the differences between personalization campaigns and experiments, not every denominator works with every numerator.

Numerators Experiment type Possible denominators
Per visitor Per session Per conversion
Unique conversions and total conversions Optimizely Web Personalization campaign   Yes  
Unique conversions and total conversions Experiment Yes    
Bounce rates and exits rates Any   Yes *  
Total revenue and total value Optimizely Web Personalization campaign   Yes Yes
Total revenue and total value Experiment Yes   Yes

When you run an experiment with many variations and metrics, there is a greater chance that some of them will give false positive results. The Optimizely Web Experimentation Stats Engine uses false discovery rate control to address this issue and reduce your chance of making an incorrect business decision or implementing a false positive among conclusive results. To learn how Stats Engine prioritizes primary and secondary metrics and monitoring goals, see Stats Engine approach to metrics and goals.