Merchandizing Audience Optimization

Zeta brings the power of our leading MarTech software directly to your Snowflake instance. Using Streamlit and Snowpark, the Zeta Merchandizing Optimization Application helps you to generate relevant target audience for your products.
Whenever you need to identify suitable audience for the products you have in stock or would like to promote, or you need to decide how big discount (if any) should you give to a particular customer to not cannibalize their revenue, our app can help you with that.

In Zeta Merchandizing Optimization Application, an advanced technique is used that utilizes statistical algorithms to recommend relevant customers for any of your products. For this process it needs just a single data source at its input, containing sales transactions with product-level details. This data contains attributes such as date, order number, customer ID, product ID, price, quantity.

The structure of the source table needs to be as follows:

Column names type null? primary key unique key
Invoice VARCHAR(16777216) Y N N
StockCode VARCHAR(16777216) Y N N
Description VARCHAR(16777216) Y N N
Quantity NUMBER(38,0) Y N N
InvoiceDate DATE Y N N
Price FLOAT Y N N
Customer ID FLOAT Y N N
Country VARCHAR(16777216) Y N N

The transactional dataset should be stored in a for of Snowflake table and linked via the app.

The algorithm generates product recommendations through the analysis of transactional data. The process begins with the ingestion of transactional data, followed by a customizable date range filter to refine the dataset. Users have the flexibility to select specific products for which recommendations will be generated.

To further tailor the recommendations, users can opt to organize transactions into baskets based on either Customer or Invoice. Additionally, users can set the minimum number of products that a basket size must have to be considered, and they also can define a minimum quantity threshold that a customer must have bought for a particular product to be considered in the recommendation process.

The user must also specify a minimum support threshold, that will serve as a filtering mechanism, ensuring that only associations meeting a certain level of significance contribute to the final recommendations.
After execution, the algorithm produces a list of customers with potential interest in the selected product or products. It also provides an additional list comprising past buyers of the same product, and a suggested discount tailored to each customer based on past behaviour.
Change the parameters and change the prediction, all in an easy to interpret visual dashboard. Zeta uses Snowpark UDF, Streamlit and the Native App framework to enable this merchandizing optimization capability. A preview of the application using Kaggle sample data is immediately available by specifying the table names mentioned within the relevant help.

Business Needs
Generate relevant target audience for merchandizers’ products

• Merchandizers need to identify suitable audience for the products they have in stock / would like to promote
• They also need to decide how big discount (if any) they should give to a particular customer to not cannibalize their revenue

Security
Consumer data is kept private and secure. After the app is installed, it is recommended by the provider to grant the following privileges as needed.