Next Best Action models

Algorithm name: Next best action

Applies to: B2B, B2C

Next best action algorithm helps recommend the most relevant action for every customer/ account based on their profile, behavioral and transactional history

What is Next best action algorithm?

The Next best action algorithm is a ready-to-use data science model that recommends the most relevant and impactful action for each customer. The model uses customers’ profile data, behavioral data, and their purchase information to generate recommend the best action that are most likely to drive positive outcomes and improve engagement.

Parameters of the model

To create and configure the Next best action model the following parameters must be defined:

  • Algorithm: Choose the Next best action algorithm in Unity.

    • Action catalog: You must choose the action catalog from which the actions need to be recommended

    • Top N actions: You must choose the number of actions to be recommended for every customer. The default value chosen is 5 while you can choose to populate up to 20 recommendations.

  • Queries: The queries you select generate the dataset used for both model training and scoring.

  • Inputs: Inputs are attributes from the Oracle Unity data model that the model uses during training and scoring.

  • Outputs: These are the data objects and attributes from the Unity data model that store the model's output values. You can customize the default output mappings if needed.

Model inputs

The Next best action model uses the following data. For the model to successfully run, check the sections: key input considerations, key data guidelines and best practices.

Attribute from Unity data object

Attribute name in the query

Unity Data object

Description of the attribute

Expected Data Type

Must have?

ID

MasterCustomerID

MasterCustomer

Customer identifier

String

Y

Country

Country

MasterCustomer

Helps asses regional similarities in customer preferences

String

 

Age

Age

MasterCustomer

Helps assess age based similarities in customer preferences

Int

 

Gender

Gender

MasterCustomer

Helps assess gender based similarities in customer preferences

String

 

SourceID

SourceID

Customer

Unique Identifier for the source 

String

Y

EventTS

EventTS

Event

Timestamp of when an event occurred

timestamp

Y

Type

EventType

Event

Engagement event type

String

Y

SubType

EventSubType

Event

Engagement event subtype

String

 

URL

EventURL

Event

Page URLassociated with the event

String

 

Medium

Medium

Event

Engagement channel through which an event was captured

String

 

ProductID

ProductID

Event

Product Identifier associated with the engagement event

String

 

Type

ProductType

Product

Type of product associated with the event 

String

 

SourceCategoryID

CategoryID

Category

Promotion category identifier

String

Y

Type

CategoryType

Category

Type of category (catalog) - offer for NBO and action for NBA

String

 

Name

CategoryName

Category

Promotion category name 

String

Y

ID

Category_ID

Category

Promotion category identifier

String

Y

SourceID

Category_SourceID

Category

SourceID from the Category object

String

Y

ID

PromotionID

Promotion

Unique Identifier for the promotion

String

Y

Name

PromotionName

Promotion

Promotion Name

String

 

Key considerations on model inputs

  1. Data schema checks: If an attribute is unavailable, assign a constant default value across all records. This allows the model to validate the input schema. This will not impact the model outcomes as the attribute will be excluded during feature importance analysis due to lack of data variance.

  2. Schema consistency: Ensure that the attributes used in the query exactly match the specified 'Attribute name in the query' (in the above table) to avoid schema mismatches during model execution.

  3. Non-mandatory attributes: These attributes, when available, can enhance the model’s learning and improve its predictive performance. However, if they are missing, the model will still leverage other available attributes to make predictions.

Key input data guidelines for Propensity models

  1. Catalogue requirements: Ensure the promotion catalogue (category) and promotion offers/ actions are ingested before the model training (refer documentation on ingesting catalog data for the model)

Best practices

  1. Use case first approach: Begin with a specific use case to determine the data required for solving the problem effectively.

    1. Review model data requirements with both business and data teams to ensure alignment

  2. Contextual parameters: Choose model parameters (such as lookback window) based on the business/ use case context.

  3. Query validation: Test both the training and scoring queries to validate that they return data and inspect sample records resulting from the query

Model outputs

Output values will be stored in the MasterCustomerRecommendation data object. You can review the ranks and scores for each recommended action in the ‘Rank’ and ‘Score’ attributes.

Attribute

Description

DataType

MasterCustomerID

Master customer identifier

STRING

PromotionID

Promotion identifier – Offer, Action or Product recommended

STRING

SourcePromotionID

Promotion identifier – Offer, Action or Product recommended

STRING

SourceMasterCustomerRecommendationID

Unique identifier for the recommendation object

STRING

SourceMasterCustomerID

Master customer identifier (source)

STRING

SourceID

Unique identifier for the source

STRING

Score

Recommendation score (Higher the score, better the affinity)

FLOAT

Rank

Rank for the promotions based on 'Score'

INTEGER

Create and use a Next Best Action model

To create and use a Next Best Action model, you will need to do the following:

  1. Before creating the data science model, you will need to create the catalog of actions that will be used when the model runs. Learn more about Creating Next best actions.

  2. After creating the catalog of actions, follow the steps for Creating Next Best Action models.

  3. After creating the model, follow the steps for Running training and scoring jobs.

After the model runs and creates output values, you can do the following:

data science, data science model, analyze data, create data science model, how to create a data science model, nba, next best action, next best action model