Three years ago, Johnson & Johnson (J&J) set out to apply intelligent automation (IA) to every aspect of its business. As the global COVID-19 pandemic was beginning to spread, the company, one of the world’s largest suppliers of pharmaceuticals, medical devices, and consumer packaged goods, needed to reduce costs, speed up tasks, and improve the accuracy of its core business operations.
Robotic process automation (RPA) was already gaining traction as organizations sought to apply software “robots” to automate rules-based business processes. But organizations like J&J wanted to take automation further. By combining RPA with machine learning (ML) and artificial intelligence (AI), they sought to automate more complex tasks. The opportunity led J&J’s Ajay Anand and Stephen Sorenson to place a very big bet in 2021.
“The one way to get attention in J&J from your very senior leaders is with the size of the impact that you could have,” says Anand, the pharmaceuticals’ vice president of global services strategy and transformation. “Generally, J&J prefers everything in billions.”
Anand and Sorenson, the company’s senior vice president of technology services, supply chain, data integration, and reliability engineering, proposed the creation of an enterprise-wide Intelligent Automation Council that they would chair. And they said they would deliver half a billion dollars of impact over the following three years. The team has already nearly hit that mark. Anand notes that, in a recent review, an executive committee member asked them to double that number based on the current pace.
Early intelligent automation roadblocks
Thanks to the work of the Intelligent Automation Council, J&J is now applying IA to everything from basic business processes, to chatbots that can help employees and customers, to algorithms that can monitor the company’s supply chain and help it adjust to changing conditions — like a doubling of the demand for Tylenol in the early days of the pandemic.
Stephen Sorenson, SVP of technology services, supply chain, data integration, and reliability engineering, Johnson & Johnson
Johnson & Johnson
But when Anand and Sorenson helped J&J take its first steps on its automation journey, they quickly ran into roadblocks.
“We were offshoring and using low-cost labor and trying to simplify our processes, but it was very difficult to scale and turnover was high,” Sorenson says. “We had this scenario where we were constantly retraining people and exception processes were killing us.”
It’s difficult to imagine just how many exceptions a process has until you actually execute on it or train people to do it, Sorenson explains. Exceptions can gum up even seemingly simple tasks, like sending confirmation forms. Typos, a new job title — any little thing could send those forms straight into the error queue, Sorenson says.
“We tried to automate them and what we realized was that people didn’t know their business processes as well as they thought they did,” he explains. “They knew their jobs and they could get work from point A to point Z, but if you tried to automate that, very few of the automations had an easy path to the end.”
It didn’t take long to realize that the traditional approach to mapping business processes — sitting down with employees, understanding how they go about their work, and capturing that — wasn’t going to give the automation team what they needed. To get a complete view of business processes, J&J brought in a task mining tool.
“We picked a handful of employees who were willing to partner with us in the early stages and we went through all of their privacy concerns and trained them, then we put this tool on their desktop to record the actual activity,” Anand explains. “When they were starting a specific process, they would hit record, and then we would capture it on this tool. We ended up creating the swim lane and all the documentation associated with it.”
Rather than interviewing the employees about the process up front, the team took the recordings and reviewed them with employees, asking whether there were any variations that weren’t captured that they wanted to share.
Adopting a digital-first mindset
J&J started using RPA for simple business process tasks such as moving documents, filling out spreadsheets, sending key messages, email integrations, and the like. It grew from there.
Ajay Anand, VP of global services strategy and transformation, Johnson & Johnson
Johnson & Johnson
“When we looked at all of our business processes, we were also very keen on ways in which we might be able to reimagine them with a digital-first lens,” says Anand, pointing to invoice-to-cash as a key example of the company’s new perspective. Like any company, when executing that process, J&J sometimes had errors or disputes with customers.
“By reimagining those processes with a digital-first mindset, we were able to look at things end-to-end and look for places where we are not only just able to automate, but also incorporate some intelligence,” he says. “Can we predict the customers with which we may have some disputes, and can we start taking some steps, proactively?”
By applying intelligent automation to invoice-to-cash, J&J was able to increase cash collection, reduce the error rate, and reduce the number of work hours and dollars spent to achieve the same results.
Anand explains that the core of J&J’s digital-first mindset around intelligent automation is 3E: experience, effectiveness, and efficiency. Does the automation change the experience of employees, customers, and suppliers? Does it make processes more effective and more efficient?
Success flowed from small wins
Sorenson says the team learned that the key to successful automation, as with many IT projects, was starting small, getting wins, and educating people about the possibilities.
“We had a saying, ‘Don’t try to get a home run.’ Just get on base, get the players on base, and we’ll move them around, start getting some hits. And then we’ll start getting some runs,” Sorenson says. “That really helped people think they didn’t have to worry about everything, they just needed to get these few steps automated and then we can see where we can take it from there.”
Sorenson notes that the small wins were able to help the automation team earn trust, but they also generated data that allowed them to show that the digital-first, machine-first mindset led to more accurate results.
“If you thought about it differently, you could actually automate the steps so that they were more accurate and build in detection so that you could find issues where things were failing historically, or even reconciliation steps that allowed us to confirm that things were working all along,” Sorenson says.
Pretty soon, as trust grew, the conversations were no longer about convincing stakeholders about the value of automation; they were about what else the team could do.
Anand notes that managing fears by showing examples to peers and partners was key.
“When people saw those examples, that really inspired them,” Anand says. “There was always this little fear that automation means people are going to lose their jobs. And they were able to see that it actually moved employees to more higher-order work and freed them up to do more innovation.”
Artificial Intelligence, Robotic Process Automation