Find out how you can analyze your Unless personalizations using goals and our built-in data platform.
The Insights page helps you keep track of your performance. It contains data about:
To see results on your Insights page, you have to create at least one audience and one goal (learn how to set a goal). Unless will then start collecting data and after some time you will see the results. The data is split into two groups - new visitors and returning ones.
For each goal, we measure how often it was reached by your control group (→ Original) and your targeted visitors (→ Personalized sessions). You can switch between the tabs "new visitor" and "returning visitor". By clicking on details you can see additional data:
The image above shows you that the goal "order completed" had a conversion rate of 12% for the control group and 14% across all personalizations. The goal "newsletter singup" shows the same conversion rate for both groups - this inidicates that there are no personalizations (yet) that could influence this goal.
From the dropdown you can select a goal and see how often it was reached by each audience. Every audience is made up of control group members and participants (=visitors who saw your personalizations). You can switch between the tabs "new visitor" and "returning visitor". By clicking on details you can see additional data:
The image above shows you that for the audience "FB Christmas Campaign" the goal "completed order" had a conversion rate of 20% for the control group and 25% for participants. The avg. value is 5€ higher for members who saw personalizations. Both numbers, the higher conversion rate and the increased avg. value, show that our personalizations generate more revenue.
This is caused by measuring different metrics. The Goal performance view is based on sessions while the Audience performance view is based on visitor ID. The visitor ID is a unique identifier for each of your visitors and it is persistent (until the person deletes their cookies). Within a session, you can reach a goal once (=unique per session). But, you could reach a goal once while being part of multiple audiences (e.g. "from Germany" and "Returning"). Vice-versa, within an audience, you can reach a goal once (=unique per visitor ID). This means, even if you reach a goal multiple times in multiple sessions, it's only counted once.
Example: You have a goal called "viewed pricing". Person A visits your page 3 times (=3 sessions) and views your pricing page each time. Person A belongs to two audiences ("from Germany" and "device=desktop"). This results in the following performance:
Again, this is caused by measuring different metrics. The Goal performance view is based on sessions so the avg. value is displayed per session. The Audience performance view is based on visitor ID so the avg. value is displayed per audience member. Since an audience member can only convert once, their member value will go up while the # of conversions will remain unchanged.
Example: You have a goal called "purchased". Person A made 3 purchases in 3 days (=3 sessions). Each purchase is valued at 10€. Person A belongs to one audience ("Referrer:Facebook"). This results in the following performance:
Whether a visitor is a member of the control group is decided upon their first website visit and saved in a cookie. Changing the control group percentage, changes how visitors are distributed between control and participants. However, this change ONLY applies to new visitors. Returning visitors are not reassigned. Consequently, a change in control group % can lead to an inbalance in visitor distribution. To assure correct data interpretation, focus on the tab "New visitors" and wait a few days or weeks for the numbers to realign.
Comparing goals across different analytics tools is always tricky, typically, you'll see discrepancies of ~10%. Discrepancies can usually be explained by when the tools trigger the goal event and what they count as a completed event.
Example: A "page viewed" goal can either be counted based on unique or total pageviews. Both approaches are valid but produce different results.