Using the ‘IsNull’ function directly within filters or conditions (e.g., in a Group By or Filter shape) is the most direct and maintainable approach. It allows you to define the business logic for both null and non-null scenarios within the decision flow itself, ensuring the strategy completes and provides a result, such as a conservative offer or a request for manual review, without failing or requiring manual intervention for every null value.

The core distinction is in their operation. A ‘Filter’ shape processes customers one-by-one, applying a condition to each to determine if they proceed down a specific path. A ‘Group By’ shape operates on the *set* of customers that have reached it, dividing them into mutually exclusive groups (like buckets) based on a property value, so each group can receive different treatments. Filters control flow for individuals; Group By segments populations.

Weighted-random (or stochastic) prioritization does not guarantee a fixed distribution over many runs or set a static rank. Instead, each time the strategy executes for a customer, it performs a “lottery” where each eligible proposition’s weight determines its chance of being selected as the top priority. This introduces controlled variability, which is useful for testing proposition performance or avoiding always showing the same top offer.

Contact Policies are specific rules (like Contact Frequency Limit) that are attached directly to a proposition or offer. They define the business rules for how often a customer can be contacted for that specific proposition across channels. This is the primary and most granular mechanism for enforcing “once per 14 days”-type rules. Interaction History filters are for referencing past activity, not for enforcing policy limits.

The dynamic and real-time nature of the requirement (last 7 days) calls for using a real-time calculated property. This property can be configured to pull and compute engagement metrics on-the-fly. This property can then be used in an ‘Adjust’ shape to modify a proposition’s priority or directly within a ‘Priority’ shape’s formula, allowing the strategy to react to the most current customer behavior.

The ‘Eligibility’ section within an Audience rule is essentially a filter applied to the initial ‘Population’. It refines the broader population (e.g., “All Customers”) down to the specific subset that should be evaluated by the attached strategy. For example, your population might be “Gold Members”, and the eligibility could be “Gold Members whose last login was within 30 days”.

Filtering customers out earlier in the process is a key performance optimization. Criteria placed in the Audience’s ‘Eligibility’ are applied at the very beginning, often leveraging database-level filtering. This reduces the volume of customer records that flow through the more complex, real-time logic of the strategy canvas, leading to faster execution. It’s a fundamental design principle for efficient decision strategies.

The ‘Treatment’ is a critical metadata field on a Proposition. It groups offers that are considered similar for the purpose of applying business rules. For example, all “Home Insurance Quote” propositions might share the same Treatment. Contact Policies (like frequency limits) are often applied at the Treatment level, preventing a customer from being contacted about *any* home insurance quote more than once per month, even if there are multiple specific quote propositions.

A ‘Group By’ shape is the ideal construct for branching strategy flow based on discrete values of a property (like CustomerType). It cleanly segments the audience into mutually exclusive groups (New vs. Existing) and allows you to define distinct downstream logic (different offers, priorities, etc.) for each group on the same canvas. It’s more efficient and maintainable than parallel filters for this use case.

Using the ‘IsNull’ function directly within filters or conditions (e.g., in a Group By or Filter shape) is the most direct and maintainable approach. It allows you to define the business logic for both null and non-null scenarios within the decision flow itself, ensuring the strategy completes and provides a result, such as a conservative offer or a request for manual review, without failing or requiring manual intervention for every null value.0