Beyond Traditional eCommerce Analytics: Insights to Action with ML/AI – Part 2 of 2

In the first part of the blog series “How to Turn Traditional eCommerce Analytics Into Actionable Insights - Part 1 of 2” the focus was to modernize your digital commerce analytics and achieve actionable insights. Being able to collect, analyze, and derive actionable insights - from across your digital landscape - at scale and speed is critical for businesses to understand customer needs and drive business operations.

There are a wide variety of increasingly "effort efficient" tools in the space of Machine Learning and Artificial Intelligence that can dig through an enormous volume of data and decipher patterns to predict future customer behavior.

Once you're setup for insights, it's time to make them actionable

The ability to continuously interpret and react to customer needs has become critical in providing personalized customer experiences. ML/AI is driving more and more of these daily aspects of our interactions with products and services such as:

• Which vehicle is assigned when booking on ride-sharing services
• The next show or song suggested by a streaming service
• A product recommended for you by a digital commerce website
• A faster route suggested by maps app

So, the question is how to go from insights to actions? Continuing the focus on Google Cloud Platform from our previous article, let us explore tools and features available in the space of machine learning and artificial Intelligence. We have previously discussed how to handle the "4 V's" of data (Volume, Variety, Velocity, Veracity). But there's one more 'V'. The true value of data is when the data can drive business activities in a way not manually feasible and that’s the fifth ‘V’.

Along the customer journey on a site/app (or even in-store), there are many opportunities for data-driven interactions to enhance customer experience by staying one step ahead in understanding what customers are actually looking for or want. ML/AI technology advancements makes predicting customers' needs and acting upon them much more realistic. Here are some examples where ML/AI can help with aspects of a digital commerce journey's specific stages:

• Awareness stage - customer segmentation and content type predictions
• Consideration stage - product and features recommendations
• Decision stage - relevant offers and pricing predictions
• Retention stage - customer lifetime value and retention offer predictions

Google Cloud Platform ML/AI Features

Google Cloud Platform provides an elaborate set of tools and solutions for applying machine learning & artificial intelligence in the digital commerce space. The platform provides tools to build/deploy custom models for specific industry needs but also standard ML/AI features for general use cases like recommendations, natural language processing, image processing, and more. Here is a look at how businesses can take advantage of various tools provided by the google cloud platform.

GCP ML/AI Feature	Typical Use Case
Custom ML/AI API	Choose ML/AI Framework to build/train/test/deploy custom models Recommendations Predictions Vision and more
Recommendation API	Others you may like Frequently bought together Recommended for you Recently viewed
Vision API	Enable Visual Product Search Object detection (i.e. Damaged Goods) OCR - Label reading and searching for a certain text Shop by Style - Faceting based on visually similar products
Natural Language API	Understand customer sentiments from chats, emails, etc Content classification Multi language text analysis Entity identification in Invoices, Contracts and more
Auto ML API	Inventory forecast Price predictions Customer Lifetime Value Churn Predictions
Speech API	Customer Support Call Analysis Agent Resolution Analysis Customer Grievance Analysis Multi-language Analysis

Google Cloud Platform enables you to quickly get started using their standard ML/AI features but also create complex custom ML/AI use cases as business needs evolve.

Key Steps to Setup Digital Commerce ML/AI on Google Cloud Platform

Option 1: Standard ML/AI Features

This is an excellent starting point for businesses that are venturing into the ML/AI space for the first time. With this option, businesses can take quick advantage of ML/AI features without enormous investment. Here are steps to get started.

1. Prepare: Create a Customer Data Repository for e-Commerce Analytics (more detail here)
2. Connect your data to standard ML/AI Services
3. Choose and train models based on standard ML/AI Service 
4. Test & optimize performance
5. Call the standard ML/AI Service API to create actions
6. As previously discussed here are a few APIs
1. Recommendation API
2. Vision API
3. Natural Language API
4. AutoML Tables
5. Speech API
6. Translation API
7. DialogFlow

Option 2: Google AI Platform for Custom ML/AI

For businesses that have very complex businesses models, it raises a need to build custom ML/AI models. Option 2 provides the flexibility to customize models and parameters for specific business use cases while providing access to multiple ML/AI frameworks and infrastructures for rapid development and deployment. Here are the key steps to get started.

1. Prepare: Create a Customer Data Repository for e-Commerce Analytics (more detail here)
2. Choose Deep learning VM of your choice, such as:
1. Tensorflow
2. Pytorch
3. MXNet
4. RAPIDS and more
3. Use JupyterLab to run interactive experiments with specific AI frameworks and models
4. Create and train custom models
5. Test & optimize performance
6. Deploy on Google Cloud or Hybrid Infrastructure

Google Cloud Platform Overview for Options 1 & 2

Learn more about how AAXIS Data Insights can help you leverage Google Cloud Platform for your Digital Commerce Analytics and today!