Pega Marketing Consultant flashcards

32 cards. Read the term and the definition here. The deck at the top of this page is optional.

Next-Best-Action arbitration

The process that evaluates all eligible actions for a customer and channel, scores them by propensity, value, and levers, and returns the single highest-ranked action.

Issue tree

A hierarchy of business issues, groups, and leaf issues that organizes actions; the NBA strategy references a top-level issue such as Sales.

Applicability versus suitability

Applicability is eligibility - an action failing it is excluded before scoring. Suitability is ranking of eligible actions by propensity, value, and levers.

Propensity

The modeled likelihood that a customer will accept a specific action, produced by an adaptive, scorecard, or imported predictive model.

Business lever

A run-time ranking adjustment applied by direction; a positive lever boosts an action and a negative lever suppresses it when its condition matches.

Contact policy

An engagement rule that caps how often a customer receives contacts per action or per group within a defined time period, using contact history.

Suppression rule

A condition that excludes specific customers or actions from consideration entirely, evaluated as a filter before ranking.

Final score composition

The ranking value for an action, typically a weighted combination of propensity, expected value (propensity times action value), and lever adjustments.

Decision strategy graph

A directed flow of shapes - data, strategy, decision, proposition, and import - connected from Start to Result that executes decisioning logic.

Group By shape

A strategy shape that partitions the incoming customer set by a property and runs its inner strategy once per group, enabling per-segment aggregates.

Decision shape

A shape that evaluates a decision table or decision tree; first-match mode fires one row, all-matches mode can fire several.

Proposition shape

A strategy shape that filters, ranks, and selects from a proposition data set, applying eligibility, suitability, and arbitration inside a strategy.

Aggregate shape

A shape that computes counts, sums, or averages over a group, commonly used for contact limits such as contacts in the last seven days.

Result shape

The terminal shape of a strategy that declares the output component type, such as a Page or Proposition, matching the downstream consumer's expectation.

Strategy test trace

A built-in test run against a sample customer page that shows each shape's input and output so you can locate where an action was filtered.

Sub-strategy reuse

A strategy shape that references another strategy, keeping shared arbitration logic in one maintained place instead of duplicating it across graphs.

Proxy strategy

A strategy that exposes a decision to other applications through a defined input and output contract, decoupling callers from internal logic.

Data flow

A configured pipeline that moves and transforms customer and interaction data; real-time flows run per interaction and batch flows run on schedule.

Data set

A source or sink abstraction - stream, database, or repository - that connects the decisioning layer to external systems for reading and writing.

Adaptive model

A predictive model that learns continuously from captured outcomes, updating propensity estimates as new responses arrive without manual retraining.

Scorecard model

A predictive model built from historical data with weighted predictors, refreshed on a schedule rather than learning continuously from each outcome.

Outcome capture

Recording the customer's response to an action through an outcome shape or response stream, which feeds model learning and contact history.

Proposition versus treatment

A proposition is the business offer; a treatment is the channel-specific presentation of that offer, such as an email template or web banner.

Interaction History

The store of past contacts and outcomes that feeds contact policies, suppression rules, and predictive model learning; it must be populated reliably.

Access group privileges

Role-based permissions such as Strategy Designer, Model Manager, and Proposition Manager that restrict each persona to changing only their layer.

Revision management

Rule check-in and check-out with branches and ruleset versions, used to promote strategies and models from development through staging to production.

Reference architecture layers

Source systems feed data ingestion, then decisioning, then channels, with outcomes returning as feedback to Interaction History.

Real-time versus batch decisioning

Real-time decisioning runs synchronously per interaction under a latency budget; batch decisioning runs the same strategy over an audience for outbound lists.

Decision latency budget

The maximum acceptable time for a synchronous decision call, a non-functional requirement that constrains strategy complexity and data lookups.

Delivery lifecycle

Design, build in development, test with sample data, promote through staging, and release to production with revision management and rollback.

Model drift monitoring

Tracking propensity drift and AUC over time to detect when a predictive model's performance degrades and retraining or recalibration is needed.

Strategy health metrics

Operational indicators such as fallback rate and filtered-out rate that reveal whether a strategy is returning actions or silently excluding customers.