Explore Einstein Recipes and Einstein Decisions
Learning Objectives
After completing this unit, you’ll be able to:
- Define the parts of an Einstein Recipe.
- Determine the purpose of Einstein Decisions.
Marketing Cloud Personalization + Einstein Recipes and Decisions
Marketing Cloud Personalization (MCP) provides real-time, scalable, cross-channel personalization and AI to complement Marketing Cloud Engagement’s robust customer data, audience segmentation, and engagement platform. It uses tailored interactions with customers and prospects to increase loyalty, engagement, and conversions. Einstein Recipes and Einstein Decisions are two of the major AI components of MCP.
What Is an Einstein Recipe?
With Einstein Recipes, you can create configurable algorithmic strategies to present each customer with customized product and content recommendations throughout your various channels.
Einstein Recipes consist of:
- Ingredients: Ingredients form the core algorithm in Einstein Recipes. When Personalization creates recommendations for your customers, ingredients define the parameters for considering items in your catalog. Personalization then weights those items according to your customers’ behaviors and affinities.
- Exclusions and inclusions: Criteria added to Einstein Recipes control which items are recommended to your customers. For example, you can add an exclusion to show only items from the same category as the item a customer is viewing. A customer who views a specific gloves brand only sees recommendations for other gloves, and not items from the same brand.
- Boosters: Include boosters in Einstein recipes to boost items matching a customer’s affinity (using their affinity score) in any recommendations Personalization presents.
- Variations: Use variations in Einstein Recipes to modify the recommendations that Personalization presents so that your customers see a wider array of items. For example, you can configure a variation to force a recipe to show at most two items from the same category.
MCP uses recipes to create algorithms that present each customer with their own personalized recommendations. You can use Einstein Recipes to support specific scenarios, such as cross-sells, content promotions, and trending products.
What Is Einstein Decisions?
Every time a visitor views a promotion on your website, MCP captures the visitor’s “context,” including whether they’re a returning visitor, the device they used, and other information that gives insight into that unique individual. Einstein Decisions (Premium Edition only), MCP’s machine learning approach for next best offer decisioning, uses this context to predict the expected value of showing a specific offer to a particular individual by evaluating both the chance of completion and the business value of the offer to the company. Using these predictions, Einstein Decisions then determines what promotion or promotions to show a visitor in order to achieve the highest expected value.
How Does It Work?
Einstein Decisions uses a continuous-learning contextual bandit algorithm that allows you to easily make and return a decision on the next-best offer, action, or experience to provide to a customer in real time.
Einstein Decisions uses a library of the promotions and actions that you’re looking to make a decision on, and they’re uploaded to MCP via a feed or directly within the user interface. These promotions and actions can be promotional banners, calls to action (CTAs), and headlines, among others. When they’re uploaded, they can include eligibility and validity rules to ensure that each customer sees only the offers and actions that they qualify for. These are also tagged using content zones to determine where the promotion or action is applied in a particular campaign.
In the feature engineering screen, you can configure the datasets that you wish to use for these decisions. These datasets include contextual data, behavioral data, and custom attributes and model scores that you can upload from your offline systems. One of the most powerful features of this algorithm is its ability to use contextual data, such as referral source, device type, geographic location, and even weather, to make these types of decisions for first-time completely anonymous visitors and ensure that they’re being given an optimized experience from the second that they start engaging with you.
When a campaign is built in MCP, the business user selects the content zone that they want to return a specific promotion or action for, which filters down the full list of promotions and actions to a smaller subset. This list is then cut down further for each customer based on the promotions and actions that they are eligible for using the eligibility and validity rules.
The Einstein Decisions algorithm then looks at the list of promotions and actions that are available for each customer and, using that combination of contextual and behavioral data, matches the right promotions or actions for that customer based on the business outcome that you’re optimizing for. This can be something like a click or a conversion. The system then returns that promotion or action over to the end channel for that campaign.
Recap
Einstein Recipes and Einstein Decisions are the AI backbones of Marketing Cloud Personalization. Einstein Recipes uses smart algorithms to suggest products and content tailored to each user. Einstein Decisions helps you decide on the next-best offer, action, or experience to provide to a customer in real time. These two tools take the personalization portion of Marketing Cloud Personalization to the next level.