Start with a different question

A manual personalization report uses many of the same metrics as a traditional test report, but you read them differently.

• With a traditional test, you’re asking: Which variation performed best?
• With a manual personalization, you’re asking: How is each audience performing with the experience it receives?

That distinction matters because each personalized audience sees the variation assigned to it. You’re monitoring those experiences, not running them head-to-head to find one winner.

[IMAGE: Traditional test → “Which variation won?” | Manual personalization → “How is each audience performing?”]

Read the results table

Start with the results table to see how each variation is performing against the selected goal.

[IMAGE: Manual personalization results table with audience/variation, Sessions, Conversions, and Conversion rate called out.]

The main metrics are familiar:

• Sessions: How many browsing sessions were attributed to the variation
• Conversions: How many times visitors completed the action tracked by the selected goal
• Conversion rate: The number of conversions divided by the sessions associated with that variation
• Users: The number of unique visitors who saw the variation

The report displays results for the selected date range and goal. If your personalization tracks multiple goals, the target goal appears by default, and you can switch goals to explore other outcomes.

For example, you might see:

Enterprise visitors → Enterprise variation

1,200 sessions · 96 conversions · 8% conversion rate

Small-business visitors → Small-business variation

1,800 sessions · 198 conversions · 11% conversion rate

Those numbers tell you how each personalized experience is performing with the visitors receiving it. They don’t tell you that the small-business variation is “better” than the enterprise variation — the two experiences are serving different groups.

Don’t treat the base like a control

This is one of the biggest differences between reading a personalization and reading a test.

• In a traditional test, the base and variations are competing for comparable traffic.
• In a manual personalization, visitors who match an audience receive its assigned variation. Visitors who don’t match a personalized audience receive the base.

[IMAGE: Enterprise audience → Enterprise variation | SMB audience → SMB variation | No audience match → Base.]

That means the base is a fallback experience, not a control group for your personalized audiences.

For example, imagine visitors from one paid campaign historically convert at a lower rate than the rest of your traffic. You personalize the page for that group and improve their performance — but they may still convert at a lower rate than the unmatched visitors seeing the base.

That doesn’t automatically mean the personalization isn’t working.

Instead of asking whether a personalized variation “beat the base,” read each audience in the context of the visitors it serves.

Look at performance over time

The performance chart shows how conversion rates changed over the selected date range, while the traffic allocation chart shows how much traffic each variation received.

[IMAGE: Personalization performance chart with two audience-specific variations over time.]

Use these charts to understand whether performance is:

• Relatively stable
• Changing over time
• Affected by a campaign, promotion, or other date-specific event

A single conversion-rate number gives you a snapshot. The chart helps you see the pattern behind it.

Use filters and audience insights for more context

You can use filters to narrow the report using criteria such as device type or new vs. returning visitors.

For example, you might look at how an enterprise personalization performs for:

• Desktop vs. mobile visitors
• New vs. returning visitors
• Visitors from different traffic sources

[IMAGE: Personalization results filtered to Returning visitors.]

Audience insights can also show conversion rates across different visitor subsets, helping you identify channels or segments worth investigating further. Cells with fewer than 300 sessions remain empty to reduce noise.

[IMAGE: Audience insights heatmap beneath a personalization report.]

For Enterprise sites, the account engagement table can add company-level context about visitors where that data is available.

Decide what deserves attention

A personalization report is primarily observational: it helps you monitor what’s happening with each audience and experience. As you review it, ask:

• Is the audience receiving enough traffic to give you a useful read?
• Is its performance stable or changing over time?
• Does the audience definition still reflect the visitors you intended to reach?
• Does the personalized experience still make sense for that audience?
• Do the results suggest something worth investigating or testing next?

From there, you might keep an experience that still serves its audience well, adjust the audience or variation when you have a reason to refine it, or retire a personalization that’s no longer useful.

The important thing is not to turn the report into a test it wasn’t designed to be. Use it to understand the audience you chose to personalize for — and let those observations shape what you explore next.

Ready to keep going?

You now know how to read a manual personalization report, interpret each audience in context, and avoid treating the base like a test control.

Next, you’ll learn how to interpret AI Optimize results, where variation delivery keeps adapting and a new signal — Strength — helps you understand what the system is learning.