This article is relevant only to the Restaurants vertical
A variation is a version of content, layout, or experience tested against the default (control) in an A/B test to determine which performs best. In a restaurant section, variations might include:
- Personalized recommendations
- Unique banner placements or calls to action
Variations are displayed based on targeting criteria to drive outcomes like increased engagement or higher order values.
Algorithm options
To optimize recommendations, select the algorithm that best fits your campaign objective:
-
Take Rate (TR): Prioritizes products with the highest likelihood of being selected. Products with better recommendation-to-purchase ratios score higher.
For example, a product shown 7 times and purchased 6 times ranks higher.
- Check Size (CS): Surfaces products that lead to larger baskets, based on historical purchase behavior.
- DL/ML Models: Proprietary data learning and machine learning algorithms developed by the DS team for advanced performance.
Step 1: Define widgets
In the Variables tab, select the widget types to appear in this variation
Note: Restaurant vertical does not include templates.
Step 2: Add Recommendations
- For each widget, click Add Recommendations.
- Choose the algorithm for each user type (identified/unidentified)
To learn more about algorithms, click here.
Step 3: Add Features
- Define product characteristics or custom inputs to refine recommendation logic
To learn more about features, click here.
Step 4: Set Business Rules
- Define targeting rules and configure slot allocation (how products are distributed across recommendation slots)
Repeat Steps 1–4 for each widget in the variation. To add a widget for identified users, click the + next to the selected algorithm.
Step 5: Validate Configuration
- Confirm your variation settings result in valid recommendations
Step 6: Allocate Traffic
- Distribute traffic between control and test variations
To learn more about the control group and test variation, click the links.
Step 7: Define Targeting
Targeting
| WHO | |
|---|---|
| Audience |
Which segment of the population with see the recommendations. For example: All Users Pre-defined 'identified' users |
| WHERE | |
|---|---|
| Screen |
Which stores will run this campaign. For example: Select All screens Specific screens |
| WHEN | |
|---|---|
| Date | Select dates or date ranges (by user or site timezone) |
| Day of the Week | Select specific days (such as Saturday or Sunday). Useful for weekday vs. weekend experiences. |
| Time of Day |
Set specific time ranges throughout the day (by the user or site timezone) Note: Dynamic Yield does not adjust the time based on Daylight Savings Time. |
Step 8: Activate and Publish
- Save the variation and set it to Active
- Click Publish when you're ready to launch