Einstein Prediction Builder

Segment

A subset of your dataset that includes filtering in certain records, defined by filter conditions (for example, adding a filter to include customers who do not have automatic payments). Note: Defining a segment is optional but lets you focus on specific records.

Einstein Engagement Frequency

AI predicts the right number of communications to send without going overboard.

Automation Bias

Automation bias imposes a system's values on others

A marketing team is trying to improve their messaging strategy. Which outcome would be the best to try predicting? A.How often mailed catalogs will be immediately trashed B.How likely a marketing email will be opened by an age group C.How amusing readers will find their tweets D.How likely it will snow during their Chicago networking event

B

Sales Cloud Einstein

Boost win rates by prioritizing leads and opportunities most likely to convert. Discover pipeline trends and take action by analyzing sales cycles with prepackaged best practices. Maximize time spent selling by automating data capture.

Confirmation Bias

Confirmation bias labels data based on preconceived ideas.

Example set: "No" Examples

Data examples that have a "no" value for what you are predicting (for example, a customer paying their invoice on time). This is required for a yes/no prediction but not a numerical prediction. Note: A "no" value isn't necessarily a negative outcome.

Determining how to use Einstein

Do I want to predict the answer to a yes or no question? (Binary Classification)Is this zip code a good opportunity for my business?Will this customer attrit?Does a new employee require a particular type of training?Will a flight arrive on time?Will a customer miss a payment? Do I want to predict an amount? (Regression—in Beta)For what price can we sell this home for?

Einstein Language

Einstein Sentiment and Einstein Intent. Together, these APIs harness and make sense out of unstructured data from text to help better understand your customers.

Types of Predictions from Einstein

Field Filter

When using Data Checker, there aren't enough records to build a prediction.

First, make sure you have enough records, minimum of 400, when selecting your object to predict in the setup flow. Then after inputting yes/no examples, make sure the minimums are reached. There are a couple of situations you can encounter. 1. The object doesn't have enough data so you will be stuck. 2. You have enough data but there's an issue with your filters.

Type 1 vs. Type 2 Error

If the system predicts that the applicant will be able to repay the loan but they don't, it's a false positive, or type 1 error. If the system predicts the applicant won't be able to repay the loan but they do, that's a false negative, or type 2 error.

Fields to Include

In short, include as much as you can. Use your knowledge of the business to choose which data is relevant. But remember, the more data you use in Einstein Prediction Builder, the better prediction it can be.

Commerce Cloud Einstein

Increase revenue by showing shoppers the best products for them, and eliminate the time-consuming activity of manually merchandising each individual page. Create highly visual dashboards to get a snapshot of your customer's buying patterns and use these dashboards to power up your merchandising. Personalize the explicit search (search via the search box), implicit search (browsing in the storefront catalog), and category pages for every shopper, saving your customers time and bringing your business more revenue.

Name the 7 steps to the process of building a numeric prediction

Name your prediction Select an object to predict Select the type of question for your prediction select a field to predict choose fields Einstein should use as examples Choose fields for Einstein to base your prediction on Name the custom field that stores your results

Numeric Predictions

Numeric predictions often power predictive forecasting solutions (for example, "How much revenue will this new customer bring in?"), but they are also used in other contexts like customer service (for example, "How many days will it take us to resolve this customer's issue?"). Numeric predictions also use your historical data to arrive at these numbers.

Score (result)

Predicted value on a record in your dataset. The score for a yes/no prediction, is the probability of your prediction being true. The score for a numerical prediction is the predicted number (that is, the predicted price of a house). It appears in a custom field that Einstein adds to the records.

high-level recap of the Einstein Prediction Builder process

Step 1—Define Your Use Case. Step 2—Identify the Data That Supports Your Use Case. Step 3—Create Your Prediction. Step 4—Review, Iterate, Enable, and Monitor Your Prediction. Step 5—Build the Prediction into Your Business Workflow. Step 6—Measure Success and Iterate.

Dataset

The overall set of data; the set of records on the object you're predicting (for example, the records from the Invoice object).

What should be done if data lives in multiple objects?

Use formula fields or roll-up summary fields & Create a new custom object

Einstein Voice Bots

With Einstein Voice Bots, your customers can interact with your brand with their voice.

Yes-and-No Predictions

Yes-and-no predictions allow you to answer questions like, "Is this a good lead for my business?" or "Will this prospect open my email?" AI helps you answer these questions by scanning historical data you've stored in your system. Yes-and-no predictions generally come in the form of a probability (for example, "Mary Smith has a 67% chance of opening this type of email). But sometimes probabilities are converted into scores.

Einstein Prediction Builder

a simple point-click wizard that allows you to make custom predictions on your non-encrypted Salesforce data, fast. You can create predictions for any part of your business—across sales, service, marketing, commerce, IT, finance, and even HR—with clicks, not code.

Survival or Survivorship Bias

algorithm focuses on the results of those were selected, or who survived a certain process, at the expense of those who were excluded

Einstein Bots

allow you to build a smart assistant into your "customers" favorite channels like chat, messaging or voice. Einstein Bots use Natural Language Processing (NLP) to provide instant help for customers by answering common questions or gathering the right information to handoff the conversation seamlessly to the right agent for more complex questions or cases.

Einstein Next Best Action

allows you to use rules-based and predictive models to provide anyone in your business with intelligent, contextual recommendations and offers. Actions are delivered at the moment of maximum impact—surfacing insights directly within Salesforce.

How does bias enter the system

assumptions - making assumptions about what, who, and how the build should work training data: bias in the dataset model; factors you use to build the model human intervenction (or lack therof): editing training data

AutoML

automated machine learning AutoML's data cleansing sifts through your data and detects these errors and either automatically fixes them or flags them to be fixed. AutoML also includes feature engineering, which automatically combs through your data and begins identifying the most significant features for buying the item, so you don't have to. AutoML uses automated model selection to build a unique predictive model that weighs the significance of each feature. The higher the weight—relative to the other weights—the more significant the feature is for predicting propensity to buy.

Einstein products

bots voice predeiction builder NBA Discovery Vision & Language

Predictive forecasting

can flag opportunities for review if the expected revenue entered by their sales reps doesn't match closely with predicted revenue based on historical data.

types of fields prediction builder can make predictions for

checkbox specially constructed formula fields numeric

manage risks of bias

conduct permortems identify excluded or overrepresented factors in your dataset regularly evaluate your training data

Interaction Bias

create interaction bias when they interact with or intentionally try to influence AI systems and create biased results.

measurement bias

data are incorrectly labeled or categorized or oversimplified

3 steps for starting right with AI

decide what to predict get historical data in order turn predictions to action

What 4 things does einstein allow you to do

discover insights that bring new clarity about your customers predict outcomes so your users can make decisions with confidence recommend the best actions to make the most out of every engagement automate routine tasks so your users can focus on customer success

Einstein Voice

enables all users to talk to Salesforce from any device. Einstein Voice is broken down into two buckets: enabling your organization (Einstein Voice Assistant), and enabling your customers (Einstein Voice Bots), with a smart assistant they can talk to.

Einstein Object Detection

extracts and contextualizes objects in images

Send Time Optimization

helps predicts the best time to send a communication for the highest response rate, specific to each person

Societal Bias

reproduces the results of past prejudice toward historically marginalized groups.

Einstein OOTB Applications

sales cloud service cloud marketing cloud commerce cloud

4 Main Ingredients of AI

yes-and-no predictions, numeric predictions, classifications, and recommendations.

Min # of records to predict

1

min # of records for t/F values

100 per value

max fields to be included for predictions outome

1k

How long to get results?

30 minutes to 24 hours

min # of records for example set

400

min # of records for overall daatset or segments

400

What types of objects does Einstein Prediction Builder support? A.All custom objects and some standard objects B.Only custom objects C.Only standard objects D.Only some standard objects

A

Marketing Cloud Einstein helps marketers reduce handle time by collecting and qualifying customer info for seamless agent handoff. A.True B.False

B

Classifications

Classifications frequently use "deep learning" capabilities to operate on unstructured data like free text or images. The idea behind classification is to extract useful information from unstructured data and answer questions like, "How many soda cans are in this picture?" It can even take a statement like, "I'd like to buy another pair of the same shoes I bought last time," and use that to kick off a workflow that can look up the last shoe order and place the same pair of shoes in their online shopping cart.

How Can Einstein Specifically Benefit My Business?

In IT, Einstein helps build intelligent apps, business processes, and workflows for every function and industry. In Sales, Einstein helps guide reps to the best leads and opportunities so they increase conversion rates, and close more deals. In Service, Einstein helps customers find answers instantly on their channels of choice and helps agents resolve cases faster by triaging cases and recommending the right articles. In Marketing, Einstein helps marketers send the right content, to the right customer, at the right time, on the right channel, thus increasing customer engagement. And for Commerce, Einstein helps retailers recommend the best product to each customer, at the right time, boosting revenue.

Einstein Platform

powerful tools that allow admins and developers to build a customized smart assistant for their business Discover insights that bring new clarity about your company's customers. Predict outcomes so your users can make decisions with confidence. Recommend the best actions to make the most out of every engagement. Automate routine tasks so your users can focus on customer success.

Einstein Discovery

predicts outcomes without requiring your own data scientist. anyone can get the full understanding of relevant patterns on all of the data in your company, whether encrypted or not, to make predictions on customer attrition. You can have full control of the data they're putting into the predictive model and be able to dig deeper into the predictions and insights.

How are Einstein Prediction Builder and Einstein Discovery different?

Builder - Designed for everyday Salesforce admins and business users. Discovery - Great for users who want to spend more time and dig deeper into the data.

When AI makes a prediction to a yes-or-no question, the prediction generally comes in what form? A.A value of either True or False B.A value of either 1 or 0 C.A percent value between 0 and 100 D.A value of either Success or Fail

C

Where can you find your prediction quality? A.On the Setup page of Einstein Prediction Builder B.The Data Checker during the setup flow C.The Overview tab on the scorecard D.The Predictors tab on the scorecard

C

Correlation

Correlation is a number from -1 to 1 that represents the direction of the relationship between the predictor and the predicted outcome. A positive value means that the predictor and predicted outcome tend to increase together, and a negative value means that one increases while the other one decreases. So if you see a negative value, remain calm and know that it's the direction of the relationship, not necessarily a bad outcome. Just remember, a positive correlation means that the predicted outcome is more likely to happen, and negative means it's less likely to happen.

Finish this one truism of AI: "If you can't report on it,..." A."you haven't tried hard enough." B."just report on something else." C."it probably doesn't matter." D."you can't predict it."

D

Prediction Quality graph

gives an estimate of how accurate your prediction is expected to be. The Top Predictors are the top-five field values with the strongest impact on your prediction. Remember, each predictor is made up of a field and a value. Each predictor has an impact score, which is a number between 0 to 1 that represents the strength of the relationship between the predictor and the predicted outcome. So the higher the impact, the bigger the influence of the field on the predicted outcome.

Service Cloud Einstein

Accelerate case resolution by automatically predicting and populating fields on incoming cases to save time and reduce repetitive tasks. Increase call deflection by resolving routine customer requests on real-time digital channels like web and mobile chat or mobile messaging. Reduce handle time by collecting and qualifying customer info for seamless agent handoff. Solve issues faster by giving your agents intelligent, in-context conversation suggestions and knowledge recommendations.

What is an example set? A.Set of attributes and characteristics that define your example data B.A set of existing data that you give Einstein to help it learn C.New data that you make the prediction on D.The starting point for every machine-learning model

B

What is one thing that Einstein Engagement Frequency is designed to help avoid? A.Sending emails that take too long to read B.Annoying customers with too many emails C.Sending emails to the wrong address D.Waiting too long between lead creation and lead followup

B

What's the difference between a dataset and a segment? A.The dataset plus the segment is the overall set of data. B.The dataset is the overall set of data, while a segment is a subset of your dataset that includes filtering in certain records. C.Defining a segment is mandatory during the Einstein Prediction Builder process. D.They are the same thing.

B

Which AI tool is especially helpful to customers who like to help themselves with support issues? A.Classification of incoming support emails B.Chatbots C.Personalized ecommerce sites D.Product recommendations

B

How can sales teams benefit from Einstein? A.Increase revenue by showing customers the best product to buy, at the right time B errorReduce handle time by collecting and qualifying customer info for seamless agent handoff C.Create personalized, 1:1 messages and content based on consumer preferences and intent D.Boost win rates by prioritizing leads and opportunities most likely to convert E.Increase call deflection by resolving routine customer requests

D

How does the Einstein Platform help admins and developers? A.By allowing them to store data in the cloud B.By allowing them to set up accounts with permissions C.By helping them schedule appointments for their teams D.By helping them build a smart assistant customized to their business E.By helping agents route and resolve cases faster so they can increase customer satisfaction

D

What is a reason to exclude fields from your prediction? A.Hindsight bias B.Legal concerns C.Too many fields D.A and B

D

What's a common issue you can face when using Einstein Prediction Builder? A.When reviewing the scorecard, prediction quality is too low. B.When using Data Checker, there are too many records to build a prediction. C.When your prediction is enabled, you're not getting any scores. D.A and C

D

Which component of AutoML sifts through your structured data and fixes it up before you start? A.Natural Language Processing B.Feature Engineering C.Computer Vision D.Data Cleansing E.Automated Model Selection

D

While not technically a component of AI, which part of an AI solution is key to making use of AI data and insights? A.Numeric predictions B.Classifications C.Recommendations D.Workflow and rules

D

Example set: "Yes" Examples

Data examples that have a "yes" value for what you are predicting (for example, a customer paying their invoice late). This is required for a yes/no prediction but not a numerical prediction. Note: A "yes" value isn't necessarily always a favorable or positive outcome.

Example set

Data from the past that Einstein learns from and uses to build the prediction.

Getting started with AI

Decide what to predict. Get historical data in order. Turn predictions to action.

Einstein Vision and Language

Einstein Vision and Language are a set of APIs and services for Salesforce developers to use to add deep-learning capabilities to any application, ultimately allowing end users to classify images and extract meaning from text. Einstein Object Detection and Einstein Image Classification

Can't find scores or prediction values.

Enable your prediction—make sure your prediction is enabled. If it's enabled, clone the prediction and use Data Checker to make sure you have records to score. If you have 0 records to score, this means that you used all your records as examples.

Fields to Exclude

In general, more data is better, but there are exceptions. For example, exclude fields that can introduce hindsight bias. Hindsight bias happens when a field is used as a predictor whose value can only be known after the predicted event occurs. Einstein automatically detects those but can miss some, so it's always good to look through the fields yourself. There are other reasons to exclude fields, such as ethical and legal concerns.

Marketing Cloud Einstein

Know your audience more deeply by uncovering consumer insights and making predictions. Engage more effectively by suggesting when and on which channels to reach out to customers. Create personalized 1:1 messages and content based on consumer preferences and intent. Be more productive by streamlining marketing operations.

Select the characteristics that apply to both Einstein Prediction Builder and Einstein Discovery. Predictions Top predictive factors Explainability & diagnosis Improvements Data Exploration

Predictions Top preicitve factors

Recommendations

Recommendations are key when you have a large set of items that you'd like to recommend to users. Many ecommerce websites apply recommendation strategies to products; they can detect that people who bought a specific pair of shoes also often order a certain pair of socks. When a user puts those shoes in their cart, AI automatically recommends the same socks.

Prediction set

The set of records for which Einstein predicts values.By default, Einstein predicts values for all records that are not in the example set. With the prediction set, you can change this to other records.

When using Data Checker, there are intersections in the data records.

This can happen if a record ends up in both the Yes and No set; it is considered as a Yes example. Try removing filters that can create intersections, or change the Meet All Conditions box to another option.

When using Data Checker, there aren't enough yes/no examples.

This can happen if the filtered conditions are narrowed down too much. Try adding the basic, fundamental filter first and then check the data. If there's enough, continue adding filters.

When reviewing the scorecard, prediction quality is too low.

This can happen when too many fields are removed from the example set or your object doesn't have enough relevant fields. There are not enough fields to make a strong prediction. If you're only using a subset of fields in your example set, try including more fields and run the prediction again.

When your prediction is enabled and you're not getting any scores.

This can occur if your prediction scores don't appear after enabling the prediction. It can be an issue when defining your prediction; that is, all records were in the example set or defining the yes/no examples. Simply go back through your set up flow and correctly filter your prediction. Here are actionable steps to take with screenshots: Troubleshoot Empty Predictions.

Clustering

This type of AI ingredient gathers insights from your data that you may not otherwise have noticed. For example, if you are a clothing vendor, AI might learn that both rural older men and urban twentysomethings like to buy a certain type of sweater.

Einstein Voice Assistant

Using Einstein Voice Assistant, you can enable anyone in your organization to talk to Salesforce.

Smart Assistants:

Voice input: Yep. Your savvy assistant has made it its full-time job to actually listen to you when you speak. Natural language understanding (subset of natural language processing): It's one thing for someone to listen to you, and another to really hear what you're saying. This technology means your assistant has the ability to truly understand what you're saying when you speak and to respond accordingly. Voice output (natural language generation): There's nothing like the art of conversation. So appreciate the fact that this technology enables your assistant to speak information back to you, as opposed to displaying it on a screen for you to read. (Bonus: It gives your eyes a break from screens!) Intelligent interpretation: Anticipating your needs is key. This technology can figure out what it is you want next, based on context, past behaviors, and data (for example, ordering sushi every Tuesday night online, Google surfaces ads around that time next week for sushi). Agency: We're always looking for an overachiever. This allows your assistant to take action on things you didn't necessarily request. A good example: Assistants schedule follow-up meetings based on unfinished action items on meeting notes post-meeting.

Association Bias

When data are labeled according to stereotypes, it results in association bias.

When reviewing the scorecard, prediction quality is too good to be true (>95%).

When the prediction quality is too high, we usually suspect hindsight bias. Einstein detects and removes any fields prone to hindsight bias, but it's always good to double-check your dataset for other fields that should be removed. As a reminder, hindsight bias happens when a field is used by a predictor whose value can only be known after the predicted event occurs. Try excluding fields that aren't relevant for the prediction.