Salesforce Commerce Cloud (erstwhile Demandware) has introduced a new powerful tool called “Einstein Commerce Insights” that extends the foundation of Commerce Cloud with powerful insights and capabilities, giving retailers valuable information about consumer’s shopping habits.
Salesforce Einstein, which is embedded in Commerce Cloud, empowers retailers and brands to create true one-to-one personalized shopping experiences – beginning with a shoppers’ very first click. Einstein enables data-based decisions and grows smarter over time, so you no longer need to rely on broad, demographic data that is largely assumption-based and provides a very limited view of your customer’s true shopping behaviour.
As per various blogs published by Salesforce, they are already seeing successful implementations of these features in ecommerce cloud and a few success stories with DW Einstein:
- One of the features offered by SalesForce Einstein, “Product Recommendations” is already implemented and usedby more than 100 brands on over 350 sites globally. All these brands witness that this particular feature increased their revenue between 7% and 16% per site visitor
- Black Diamond Sees 15% Revenue Boost With Salesforce Einstein’s Product Recommendations feature.
- Brand ‘Icebreaker’ which is implemented on SalesForce Commerce Cloud found that its shoppers clicked on Salesforce Einstein Product Recommendations more often (40%) which leads to 28% more revenue and an 11% overall increase in average order value.
Salesforce Einstein Key Features and Aspire’s Approach to Test Automation
Why Test Automation for Salesforce Einstein:
From a QA standpoint, we need to make sure that SF Einstein feature blends well with the product change based on season sales. For example, during every Christmas, Fall, Spring, Summer season – products are supposed to get renewed on eCommerce sites and we need to test that these changes in cloud configuration manager are not impacting the recommendations. To make that job easy and save manual effort, we can take help from test automation.
Our automation framework is programmed to dynamically validate the product variation / menu variation / sub-menu variation in the Business Manager and alter the expected results in the automated test scripts which give you a maintenance free code to test Salesforce Einstein.
This tailored automation approach can help the QA team to test the Einstein features with different variations of the product to validate the end result during UAT phase and in QA environment before deploying to production.
Einstein Product Recommendations:
Salesforce Einstein Product Recommendations uses machine learning to power one on one personalized shopping experiences throughout the site. Few features of Einstein Product Recommendations,
- Provide data-driven product recommendations to both Members and Guest Users
- Automated process on finding product recommendation tailored to each web page
Our approach to test Einstein Product Recommendation:
- Since this feature is about data-driven product recommendations, Aspire has also taken a dynamic validation approach to automate it.
- Instead of hard coding the expected result on each page under recommendation, we have made the expected result – validation dynamic by making our test script going to each parent menu and its sub category menus and collect the required validation/expected results data
- For instance, we have constructed our automation approach to collect the product data from Business Manager; say (Boots->Shoes)
- Then navigate to the ‘Shoes’ menu and verify recommendation in accordance with the product value that was collected from Business Manager
- Place order, clickstream for both member and guest. Verify if recommendation matches the actions performed
- Above TA strategy not only helps QA for their quality verification but also gives the visibility to Business team
- BA team can upload the products based on upcoming season on their own EVN and run our TA Suite to make sure recommendations are met with the BA expectation
- If it is not met, required change can be made either in Cloud Configuration manager or to their product category.
- Let’s quickly see how above strategy works with different types of users: new user/existing user/guest user.
For existing user: Product Recommendations for existing user with minimum of 1+ orders. Our TA framework collects order history / wish list and assesses the expected recommendation on every possible page.
For new user and Guest User: Our TA framework assesses the possible recommendation through clickstream and orders that may be placed on the session.
Approach for above intelligent service through Test Automation:
Einstein Predictive Sort
Connect customers to products with tailored product sorting.
- Provide the best site experience by automatically personalizing search/category pages for each shopper—anonymous or authenticated
- Save time by enabling sort personalization within your existing business tools
- Drive conversions by showing shoppers what they want, especially in micro-moments on mobile devices
Our approach to test Einstein Product Recommendation:
- This is an auto-personalization (in terms of sorting) Search/Category pages for each individual.
- We make our TA framework look for specific products to be on top of our search page. Expected products /category will be fed to TA dynamically through monitoring a number of parameters as specified below
- Our TA framework not only validates this feature on Desktop Browser but also effectively cross verifies it in Mobile devices
- Personalized product assortments will be validated for all shoppers, whether anonymous or registered through our TA Framework
Einstein Commerce Insights
Unlike the other two features, this feature is specifically designed for the Configuration manager. This feature interprets purchasing behaviour with a powerful shopping basket analysis dashboard.
- Understand how customers are buying products with easy visual tools that require zero training
- Plan better shopping experiences with metrics around product-specific sales and top co-purchase categories
- Maximize conversions by optimizing product bundles, sets, deals, and merchandising
- Develop meaningful marketing campaign strategies around each customer’s shopping habits
This feature is specifically designed by SF for Configuration Manager to get reports on sales insights like getting the report visuals for the sale of a particular item in a specific timeline. So, this feature does not fall under the TA Scope.
- Demandware Predictive Intelligence Test Automation Approach - November 14, 2017