We’ve set up Einstein on our developers org. Per the requirements in the readiness report, we’ve created 1000 leads and converted 120 of them to contacts, accounts and opportunities. When we start the process of lead scoring, it says it takes 24 hours to process , but after 24 hours it fails with a message “Modeling Failed - We are unable to set up Einstein Lead Scoring for your org at this time”. We tried again and again and it always fails.
Opportunity scoring is stuck with a message “After Einstein Opportunity Scoring is enabled, it can take around 48 hours before the initial processing is complete and users start seeing scores.” – It’s always the same message and the status never changes.
We checked and re-checked our settings based on the following help pages and it seems like we should have satisfied all requirements:
https://help.salesforce.com/articleView?id=einstein_sales_considerations.htm&type=5 (https://help.salesforce.com/articleView?id=einstein_sales_considerations.htm&type=5" style="color:#.0563c1; text-decoration:underline)
https://help.salesforce.com/articleView?id=einstein_sales_setup_enable_lead_insights.htm&type=5 (https://help.salesforce.com/articleView?id=einstein_sales_setup_enable_lead_insights.htm&type=5" style="color:#.0563c1; text-decoration:underline)
https://help.salesforce.com/articleView?id=einstein_sales_setup_enable_els_errors.htm&type=5 (https://help.salesforce.com/articleView?id=einstein_sales_setup_enable_els_errors.htm&type=5" style="color:#.0563c1; text-decoration:underline)
Does anyone have any idea what “Modeling Failed” means? Is there anything else that Einstein is looking for in order to generate scoring for leads and opportunities?
Thanks for your help!
8 respuestas
Einstein Lead/opportunity Scoring requires a full sandbox with sufficient historical data from a production org. Developer Edition orgs and scratch orgs do not have the data history required to build a scoring model. So in summary you will be able to test it only in the Full copy sandbox and that to it will only give the accurate scoring based on the historical data that you have. In short you can test the functionality how will it work but you should not compare it to your production
