E-Commerce Insights

Optimizing and planning e-commerce operations

Purpose

Our client has a large collection of sales data points containing order confirmations and credit card transactions. The amount of data points is in the order of 10 million.

The client wanted to achieve the following:

  • Understand seasonility pattern in sales over a whole year
  • Better predict outcome of new product launches
  • Ability to plan product orders in anticipation of future sales

Approach

Data Kernel first provisioned a large data warehouse based on Amazon RedShift. All existing data points were loaded into the data warehouse and a pipeline was created to ensure future data points will be automatically ingested.

The schema of the data warehouse immediately allowed to collect data for prediction model building.

Two prediction models were created.

A sales pattern seasonility model allows the client to do better revenue prediction. It uses data points retrieved from the data warehouse to build a new model with fresh data every day. The model allows the client to exactly see which article number will have approximately how many sales over the next 3 months.

Furthermore, a product launch model then allowed the client to understand, how new products will perform after the launch compared to existing products that have already been launched.

Outcome

By being able to accurately predict sales ahead of time, the client was able to better allocate shipping resources and keep inventory small.

The increased ability to understand product launches helped the client shape their product development and marketing efforts.

Overall, the client’s revenue was increased while keeping down unnecessary cost such as inventory space wastage.

Related Work

Cloud Cost Optimization

Predictive cost modelling for mission critical infrastructure

IoT Analytics

Monitoring and planning IoT network

Want to learn more?

Talk to our experts