A robotic process automation rollout in the energy management sector shows even an ambitious RPA journey begins with some small steps.
Schneider Electric is a large enterprise. The company, headquartered in France, generated €25.7 billion (about $28.5 billion) in 2018 revenue, employs 130,000 people and does business in more than 100 countries. Elizabeth Hackenson, CIO at Schneider Electric, said the company's RPA efforts got underway in earnest in mid-2018. Pilot projects paved the way for a global initiative in 2019.
"What we like to do with new technology such as this is just go into a couple of pilots," she said. "Those pilots were successful, so what we did early this year was create a global team."
Schneider Electric's business spans a range of fields, including IoT, building automation, industrial automation, and data center infrastructure such as cooling and power distribution systems. The company's stated goal is the "digital transformation of energy management and automation."
Where the bots are
Schneider Electric's RPA journey is now part of that transformative vision. The company's initial pilots paved the way for the 170 bots in production as of July 1. The majority of those RPA instances -- a bit more than 100 -- are in China.
This article is part of
"They have been more advanced," Hackenson said of the Chinese RPA adopters, citing a regional CIO who was able to rapidly identify use cases and competing technology priorities in the company's other geographic regions.
Avoiding duplicative bots
Schneider Electric's RPA team serves as the company's global entry point for software bots.
One of the team's roles is to minimize duplicated effort as requests for bot development roll in from the business side. "So, if we have a call center in Europe put in a request for a use case and another one comes in from the U.S. for something very similar, the global team knows we have already built this," Hackenson said. Without the global team, RPA activity would exist in isolation within business groups, hampering cross-company visibility.
If bots live in silos, "we don't know what is going on and we can't leverage what has been created," she added.
Schneider Electric currently uses RPA software from Blue Prism and UiPath. Hackenson said the platforms are similar, but the company continues to compare them. Thus far, bots have been deployed in customer care centers, finance, HR and supply chain.
Hackenson said she had hoped to have more than 220 software bots in operation by July 1. One issue for the shortfall has been rejected use cases. Schneider Electric's global RPA team, which resides within the company's IT department, reviews RPA use cases submitted from business leaders and has primary responsibility for bot creation. The team has been rejecting more use cases than it accepts, she said.
"We just don't want to automate siloed processes with RPA," Hackenson said. "We want to cut across a process end to end."
Elizabeth Hackenson CIO, Schneider Electric
RPA use cases may also be turned down if the processes they seek to automate are fundamentally broken. And others may fail to offer enough transformative potential.
The RPA team, however, has begun training the business side on appropriate RPA use cases and how best to take advantage of the technology.
"Since we have done the training, we have seen a big increase in use cases now being accepted," Hackenson said. "We still believe there is a learning curve on RPA."
Another issue has been setting expectations for software bot performance. In some cases, the approved RPA use cases have fallen short of the envisioned objective, which has caused the RPA team to pause and revisit some initiatives. She cited one example in which a customer call center bot was expected to take five full-time equivalents (FTEs) off inbound ordering tasks, allowing the employees to pursue higher-value activities. The initial implementation of the bot, however, only freed up two FTEs.
"We are learning it is an iterative process," Hackenson said. The organization had to rethink its labor-reduction expectations and assess whether more tasks needed to be automated to reach the objective, she noted.
Accelerating the RPA journey
The RPA team's goal is to rapidly deliver bots to the business process owners. At this point, bots take about four weeks to develop, but Hackenson said the objective is to shrink that time to two to three weeks.
"We want to move quickly and get it back to the business," she said.
Once the bot arrives on the business side, it generally takes four to eight weeks for process owners to deploy the bot and start realizing its benefits. Hackenson said the RPA team has been exploring the reasons behind the lag. One consideration is the time it takes to transfer FTEs to new assignments.
Hackenson said a change management program is likely to significantly accelerate the pace at which the business side fully uses bots. She would like to see bot acceptance take place within a two-week window.
Despite the various challenges of fielding a bot labor force, Hackenson said she remains confident of success.
"The good news is, every time we are learning," she said of the company's bot experiences.