- Optimizely X Web Experimentation
- Optimizely X Web Personalization
- Optimizely X Web Full Stack
- Optimizely X Web Mobile
THIS ARTICLE WILL HELP YOU:
- Use the new metrics builder to its fullest capacity
- Understand how the process of creating a metric has changed with the implementation of the metrics builder
In May 2017, Optimizely introduced a new way to create metrics for an experiment or campaign. In the past, metrics for an experiment were generated directly from the events available in an experiment or campaign. They weren't customizable, and you were limited to a single metric per event.
The new natural language interface for metrics allows you to:
Define metrics that more closely match your business needs, such as revenue per conversion)
View the same event as multiple metrics (for example: unique conversions per visitor and revenue per visitor)
Decide the winning direction for a metric (do you want it to increase or decrease?)
If you haven't already, be sure to check out the step-by-step guide to creating metrics in the metrics builder.
Before the metrics builder, Optimizely defined revenue metrics differently. For one thing, it was global in scope and counted across all events. This meant you could not track revenue-related metrics for individual events, and that you were limited to a single revenue metric for each experiment. The overall revenue metric was also not configurable.
The metrics builder gives you much more flexibility with revenue metrics. Instead of tracking them only at the experiment level, they are now tracked at the level of individual events. You can still apply a global revenue metric to an experiment, of course, but now you can also create several other revenue metrics for each experiment. You can also specify the winning direction and whether your metric is calculated on a per visitor, per session, or per conversion basis.
“Per conversion” is only available as a denominator in Optimizely X Web Experimentation, while “per session” is only available in Optimizely Personalization. Check out this article for more details.
The metrics builder allows you to specify the direction your metric should change in order to be declared a winner. In most cases, a winning metric is one that goes up. But sometimes, you may be interested in lowering the value of a metric (like Cancel button conversions or cart abandonment, for example) instead. By specifying whether a metric should increase or decrease, you will get a much clearer understanding of your experiment's success when you visit the Results page.
Previously, you could not change the winning direction of a metric.
The metrics builder gives you the ability to edit metrics when the experiment is running. However, this could affect the statistical significance of other metrics in your experiment or campaign. We recommend that you be careful when editing live metrics.
Unique and total conversion metrics
Previously, whenever you created a metric, both unique and total conversions were included in it and bundled together in the Results page.
With the metrics builder, these metrics are defined separately. You can either create and add both of them as individual metrics, or you can simply edit an existing metric, change the numerator from unique to total (or vice versa), and view the results again.