Salesforce wants HR departments to give its AI development, Einstein, a try. It claims its low-code tools can enable easy creation of HR AI prediction tools.
Salesforce said it's not trying to become a HR systems provider. But in detailing its recently unveiled Einstein Predictions Services, Salesforce focused on a hot button HR AI application. It said it can quickly develop an attrition risk application. Low unemployment has made reducing attrition, and the ensuing turnover costs, a priority in HR.
The do-it-yourself Einstein platform enables developers and administrators to create the HR AI tools. There's no need for hard-to-find and expensive data scientists, Salesforce claimed. It said its AI analytics can be embedded into third-party HR and ERP systems.
Its next task is to convince HR departments to use it.
Salesforce will need an HR partner
"These may be used for HR -- but unlikely," said Josh Bersin, an independent HR analyst, of the Salesforce AI tool.
Bersin pointed to Salesforce's own applications as the stronger use for its AI prediction tool. "There is data in Salesforce that can identify behaviors of top salespeople and sales leaders, and this is a big opportunity," he said.
Ben EubanksAnalyst, Lighthouse Research & Advisory
But concerning HR, "I've never met an HR department using Einstein," Bersin said. "If they want to push this, they have to market it and productize it with partners."
Salesforce has offered HR-specific capabilities and application development tools, but "we're not trying to be an HR system of record," said Allison Witherspoon, senior director of Einstein AI and analytics at the company. Salesforce does, however, want to make its AI and platform development expertise broadly available to other environments, including HR, she said.
Ben Eubanks, an analyst at Lighthouse Research & Advisory, believes Salesforce is using HR as an example for its prediction tool "because of the plentiful availability of behavioral data within the business."
"On a more practical level, every business leader, regardless of their function, worries about turnover to some degree, so it's an easy sell," Eubanks said.
Accuracy depends on data
Salesforce said prediction accuracy will vary from customer to customer. It provides the means for building the HR AI application, but it uses the customer's own data and business processes, Witherspoon said.
The Einstein system also has some anti-bias capability. It finds fields that can be, for instance, a proxy for race, such as ZIP codes, and then warns users. Excluding certain ZIP codes can create a potential for racial bias. This prediction capability is still in pilot.
There are many vendors that provide attrition prediction tools. But Witherspoon said that the Salesforce tool can fill gaps in the HR AI space, and cited Workday as an example.
"Workday doesn't provide the ability to do attrition prediction, "Witherspoon said. "Salesforce does." Workday, on the other hand, said that's not the case.
Barbry McGann, senior vice president of product management at Workday, said its flagship HR application offers machine learning-driven risk functions that enable customers "to quickly identify and understand retention risk for the entire organization or a specific department, including the number of top performers at a high risk of leaving in the next year and the projected cost to replace them, as well as risk factors unique to their organization or department."
McGann said its system can also deliver retention recommendations, such as suggesting the best career moves for employees by analyzing factors such as how long previous at-risk employees stayed in subsequent roles.