Ad
More Related Content
What's hot (20)
Similar to Einstein Analytics Prediction Builder (20)
![[NYC Scrum] 4 bad ideas about productivity... and what Agilists should do ins...](https://cdn.slidesharecdn.com/ss_thumbnails/nycscrum4badideasaboutproductivity-250320014757-d85c2fbe-thumbnail.jpg?width=560&fit=bounds)
Ad
More from rikkehovgaard (17)
Ad
Recently uploaded (20)
Einstein Analytics Prediction Builder
- 1. Einstein Prediction Builder All you need to know Colin Linsky Lead Solution Engineer, EMEA Einstein Analytics Team [email protected]
- 2. Topics for today 1. AI at Salesforce 2. Creating intelligent, data driven experiences 3. Seeing is believing – a worked example of Prediction Builder
- 3. 1. AI and Salesforce
- 4. The Fourth Industrial Revolution TODAY1970s1800s1700s 4th Industrial Revolution Intelligence 3rd Industrial Revolution Computing2nd Industrial Revolution Electricity1st Industrial Revolution Steam
- 5. Don’t just take my word for it! https://www.weforum.org/agenda/2016/01/the-fourth-industrial-revolution-what-it-means-and-how-to- “Whether consumers or businesses, customers are increasingly at the epicenter of the economy, which is all about improving how customers are served.”
- 6. Adopting AI “Many organizations have successfully launched cognitive pilots, but they haven’t had as much success rolling them out organization-wide. ”
- 7. Salesforce Einstein Augmented Intelligence for CRM world’s smartest CRM = customer data + +AI Salesforce Platform
- 8. The Enterprise Can Benefit From AI Too Iulia Creanga Sales Process Driver Schneider Electric Katrice Riley Client Success Specialist ZeroCater Suvi Silvanto Marketing Director, Service Business Kone Annika Milton Sr. Project Manager adidas Zac Otero Salesforce Lead BKD CPAs & Advisors Sales: increase conversion rates Service: increase customer satisfaction Marketing: increase customer engagement Commerce: increase revenue IT: be innovative, always 84% of enterprise execs believe AI will help them create a competitive advantage Source: “Reshaping Business with Artificial Intelligence” MIT Sloan Management Review & Boston Consulting Group
- 9. Einstein Makes our Apps and Platform SMART Store Data Make Predictions Analyze Results Take Action Rapidly Automate Trusted Cloud myEinstein Lightning Einstein Analytics Sales Cloud Einstein Service Cloud Einstein Marketing Cloud Einstein Community Cloud Einstein Commerce Cloud Einstein Einstein Prediction Builder Einstein Discovery Einstein Vision & Language CRM data Email, calendar, social data IoT, external data Einstein Apps
- 10. “Nice, you’ve built templated applications to support standard decisions but my business is unique – can I use the same technology free-form to support my own Use Cases?”
- 11. 2. Creating intelligent data driven experiences
- 12. Will this lead convert? Will this opportunity close? Will this customer stop using our services/contract? Will this customer renew a contract? Will this case escalate? Will this case be solved at First Contact? Can I upsell to this customer? Will this customer open or click an email? Will this invoice be paid in time? Will this product pass QA? “Yes or No…..”
- 13. … is the CSAT going to be? … revenue will we make with this customer? … longer before this opportunity closes? … longer before this case closes? … longer before we need to service this installation? … products will be sold this month? … times will I need to contact this Doctor? … cases are needed to resolve this issue? … times a week will orders be placed? “How much…..” “How many…..”
- 14. … is the CSAT going to be? … revenue will we make with this customer? … longer before this opportunity closes? … longer before this case closes? … longer before we need to service this installation? … products will be sold this month? … times will I need to contact this Doctor? … cases are needed to resolve this issue? … times a week will orders be placed? “How much…..” “How many…..” Step One: Define The Problem Positive Examples Records that have churned Negative Examples Records that have not churned Records that we will give a score for Records to Predict Example: “I want to predict churn, but what does it mean?” Break your records down into three or four buckets Records to Ignore Irrelevant records (Account has not purchased in the last six months) (Account have purchased in last three months) (Account that have not purchased in the last three months) (Accounts that don’t fit into the other buckets)
- 15. … is the CSAT going to be? … revenue will we make with this customer? … longer before this opportunity closes? … longer before this case closes? … longer before we need to service this installation? … products will be sold this month? … times will I need to contact this Doctor? … cases are needed to resolve this issue? … times a week will orders be placed? “How much…..” “How many…..” Step Two: Capture Churn in a Formula Field “TRUE” Records that have churned “FALSE” Records that have not churned Records that we will give a score for NULL Example: “I want to predict churn, but what does it mean?” Represent the buckets as responses from a formula field “IGNORE” Irrelevant records (Account has not purchased in the last six months) (Account have purchased in last three months) (Account that have not purchased in the last three months) (Accounts that don’t fit into the other buckets)
- 16. Transmogrifai
- 17. Decision Support in the Real World Prediction Builder (Core) Easy admin, self-sustained prediction engine and deployment Stories in Einstein Analytics (Einstein Analytics Plus) Flexible, comprehensive, open and inclusive insight generation and model building capabilities Einstein Analytics (Einstein Analytics Growth) Tools for surfacing insights right where they are needed, in the format that has the most impact, and ready to take action from * Prediction Builder + Discovery (Einstein Predictions) Building and embedding your own predictions
- 18. Einstein Analytics Interactive analytics and production-ready dashboards & reports Interactive Analytics Unlimited exploration of relationships and differences Deployable Insights Embed charts, reports, metrics, guides and many other types of insight directly into consumer and customer systems using widgets Industrial Strength Data Handling Features and functions designed to work seamlessly with customer data inside and outside of Salesforce Native CRM Insight into Action Framework See something and then take action on it right from within the insight
- 19. Einstein Analytics Stories Advanced analytics at the heart of Salesforce Automated Analytics Analyze millions of data combinations in minutes Unbiased Insights Understand what happened, why it happened, what will happen, and what to do about it Narrative Explanations Natural language-based insights and stories exported to Salesforce or Microsoft Office Recommended Actions Take action, stay on top of changes, and gain advantage right where the decisions happen speed to insight +10x *Source: Salesforce Customer Relationship Survey conducted 2014-2016 among 10,500+ customers randomly selected. Response sizes per question vary.
- 20. Prediction Builder Automated Predictive Model building and deployment Automated Analytics Wizard driven setup and ease of administration Adaptive Sensitivity Intelligent model building and control of inputs Predictions Embedded into Records Automated Model Management Model competition and automatic sensitivity to changes in data patterns Writeback process automated https://engineering.salesforce.com/open-sourcing-transmogrifai-4e5d0e098da2
- 21. 3. Finding, configuring and running Einstein Prediction Builder
- 22. Demo Wouldn’t it be great to know in advance which invoices were going to be late to settle?
- 25. Einstein Prediction Builder Accesses data from a single object at a time • All data fields must be found on the Salesforce object • Prepare data (including rollups/lookups/calculations/formulae) ahead of Model building phase • Multiple Prediction Builder models can be built on a single object using • Segmenting • Filtering • Model building record set vs scoring record set – bake that mindset into data prep! Modelling Data Tips: Operations - Basics
- 26. Einstein Prediction Builder works with all custom objects, and supports the following standard objects. Supported Objects Tips: Operations – Salesforce Objects Account Asset Campaign Case Contact ContractLineItem Entitlement Lead LiveChatTranscript Opportunity Order OrderItem Product2 Quote QuoteLineItem ServiceContract
- 27. Einstein Prediction Builder can make predictions for the following types of fields. • Checkbox • Specially constructed formula fields • Numeric Supported Data Types for Target Tips: Operations
- 28. So what does all of that mean?
- 29. o Models built by Admins and Scores consumed by end users, customers or colleagues o Admins and Business Users who know their domain and data work as a team (it’s not a “build me a model request”!) o Quick/Simple to set up and easy to maintain o Designed for a single Salesforce object per model o Targetable to segments of the records for model building and model scoring o Automatically score records daily o Automatically the models refresh monthly o Can be deployed and used in conjunction with Einstein Analytics Plus Stories models Einstein Prediction Builder Models are: Key Takeaways
- 30. Thank You First Name Last Name Title of Presenter [email protected] @twitterhandle Remember to tell us what you think in the event survey www.LondonsCalling.net/survey/
- 31. Back up slides
- 32. The Details tab on the prediction scorecard shows a list of predictors for your model. It shows information about each predictor, such as impact, correlation, and importance or weight. If your predictive model contains more than 100 predictors, you might not see them all. The scorecard shows you the top 100 predictors ranked by impact, and the top 100 ranked by correlation. If your model has at least 100 predictors, the number displayed in the scorecard is likely to be between 100 and 200. Limits Operations
- 35. Setting up a new Prediction Builder Model
- 36. Creating the Model name
- 37. Selecting the Salesforce object
- 38. Selecting the target field and modelling records
- 39. Selecting (potential) predictor fields
- 41. Creating the scoring fields
- 42. Final check
- 45. Confirmation that the model is being sent to the factory!
- 46. Awaiting news…
- 48. Model overview
- 50. Storing the predicted values
- 51. Viewing the model scores