Define interventions

Interventions are rules that trigger when user attributes meet specific conditions. They enable real-time personalization by activating experiences based on current behavior.

When an intervention triggers:

1. Signals evaluates attribute values in real time
2. If the criteria are met, the intervention activates
3. Your Web application can receive intervention notifications via the browser tracker plugin directly on the frontend
   • This uses a Server-Sent Events (SSE) connection for real-time streaming

In this section, you'll create three interventions for common ecommerce scenarios: cart abandonment, discount offers for engaged browsers, and free shipping promotions.

Import intervention classes

Add the intervention-related imports to your notebook:

from snowplow_signals import (
    RuleIntervention,
    InterventionCriterion,
    LinkAttributeKey,
)

Define intervention rules

These rule-based interventions will be triggered automatically when their criteria are met.

Each InterventionCriterion specifies:

• attribute: the attribute to evaluate, in the format group_name:attribute_name
• operator: the comparison operator (>, <, >=, <=, ==, !=)
• value: the threshold value for comparison

Cart abandonment intervention

This intervention triggers when a user has added items to their cart but hasn't completed checkout:

cart_abandonment_intervention = RuleIntervention(
    name="cart_abandonment",
    description="Show banner to users who have added at least one item to cart.",
    criteria=InterventionCriterion(
        attribute="ecom_attributes:count_add_to_cart",
        operator=">",
        value=0,
    ),
    target_attribute_keys=[
        LinkAttributeKey(name="user_id"),
        LinkAttributeKey(name="domain_userid"),
    ],
    owner="user@company.com",
)

Key components:

• criteria: triggers when count_add_to_cart > 0.
• target_attribute_keys: specifies which user identifiers to use when matching interventions. This intervention targets both the authenticated user_id and anonymous domain_userid identifiers, to reach as many users as possible.

Discount intervention

This intervention offers a discount to users who are actively browsing products:

discount_intervention = RuleIntervention(
    name="discount",
    description="Show banner to users who have viewed more than 3 products.",
    criteria=InterventionCriterion(
        attribute="ecom_attributes:count_product_views",
        operator=">",
        value=3,
    ),
    target_attribute_keys=[
        LinkAttributeKey(name="user_id"),
        LinkAttributeKey(name="domain_userid"),
    ],
    owner="user@company.com",
)

This intervention activates after a user views more than three products, indicating high purchase intent.

Free shipping intervention

This intervention encourages larger purchases by offering free shipping:

free_shipping_intervention = RuleIntervention(
    name="free_shipping",
    description="Show banner to users who plan to spend at least $100.",
    criteria=InterventionCriterion(
        attribute="ecom_attributes:total_cart_value",
        operator=">",
        value=100,
    ),
    target_attribute_keys=[
        LinkAttributeKey(name="user_id"),
        LinkAttributeKey(name="domain_userid"),
    ],
    owner="user@company.com",
)

This triggers when the cart value exceeds $100, incentivizing users to complete their purchase.

Publish the interventions

Publish all three interventions to your Signals Sandbox:

sp_signals.publish(
    [
        cart_abandonment_intervention,
        discount_intervention,
        free_shipping_intervention,
    ]
)

Once published, these interventions are active and will trigger in real time as users interact with your application.

In the next section, you'll use the demo e-shop application to see these interventions in action. The app uses the browser tracker plugin to automatically receive and display intervention banners when they trigger.