Back to Digital blog

To succeed with bots, humans need to play a bigger role

By Marty Pavlik and Saurav Tripathy

Robots are falling short of their potential. Implementation complexity and maintenance needs, among other things, are holding back robotic process automation (RPA). Here are four ways humans can help:

1. Build implementation teams with the right mix of RPA expertise

RPA can be a very powerful tool for a company to implement— “implement” being the key word. Processes automated in a few weeks can result in tremendous improvement in process quality and reduce process costs by nearly 80 percent—but only with the right capabilities and teams in place.

Leaders with realistic expectations

RPA leaders tend to overestimate the speed of RPA implementation, which leads to difficulties in effectively handling resources, managing expectations, and capturing targeted ROIs. Prudent leaders with a clear vision of the future and realistic expectations based on a deep understanding of the technology are the need of the hour. These RPA leaders are good at distinguishing quick wins from full-scale implementations, for example. This helps to keep teams motivated and allocating resources based on need.

Maintenance teams

Businesses typically turn to a third-party service provider to develop bots quickly, but fail to involve the internal employees who are charged with crucial maintenance and upgrades. As a result, those employees aren’t familiar enough with the original code, process design, or assumptions to be able to function effectively. There needs to be a deliberate and clear knowledge transfer between bot builders and maintenance teams. Costs of neglecting this run high, since more time spent on bot maintenance means fewer-than-anticipated benefits from the deployed bots.

“Translators” who can work across departments

Many organizations select members of their RPA team based on their technical prowess or domain knowledge alone. They underestimate the need for the RPA team to work with other parts of the organization. Making sure the project is in full alignment with IT requires team members to meet with IT on a regular basis. This responsibility might be best delegated to someone with deep existing relationships within the enterprise coupled with effective project- management skills rather than to someone whose only expertise is technology. Similarly, in working with the business side, the RPA team must go beyond just producing bots for the business to effectively addressing business concerns around service level agreements (SLAs) and change-management efforts. These translators also track bot performance against business goals and communicate additional requirements back to the RPA teams for a smooth transition.

2. Set up the right governance

People tend to think “it’s easy” to work with bots because they are flexible and easy to configure. But that leads to a false sense of the effort required to work on them, resulting in a large number of change requests that aren’t essential or don’t add value. Having an established governance model provides a clear process for making decisions about what—and what not— to do.

Determine RPA ownership

One of the inevitable questions that comes up during an RPA implementation is who ultimately owns the bot. The automation team believes they do, because they typically do its initial configuration and will ultimately maintain it. The business argues that it owns the bot as a resource and should decide how it operates within the business. Without a clear governance policy in place, this lack of clarity on ownership often causes delays and can result in an underperforming bot. The framework for deciding bot ownership ideally takes into account factors like process complexity, implementation, and maintenance, in addition to overall business impact.

Establish maintenance regulations

Automation teams try to move as many bots as possible into production quickly. This focus on agility sometimes results in automation teams neglecting maintenance processes and protocols. As the number of bots grows into the hundreds, there will be different versions of applications, security patches, process updates, and process changes, all of which need to be managed. Without clear protocols about who performs the maintenance and when, the sheer numbers of change requests and technical exceptions could halt the entire bot workforce. A governance model comprising IT, process owners, and other key stakeholders providing direction, guidance, and support to the automation team is essential for ensuring bot maintenance.

3. Build a strategy to use RPA where it can create value

A lot of vendors claim that RPA can solve almost any problem or improve any process. In the same spirit, overenthusiastic proponents are implementing RPA across a wide range of areas. Companies have difficulty assessing solutions and options because they don’t have a clear strategy. Before investing time and resources in building or buying RPAs, companies need a clear strategy to determine what capabilities are needed to deliver value for the business, and what role RPA has in supporting those capabilities.

Some tech leaders tend to favor RPA too much, promoting it when another technical solution might be more effective. Sometimes the automation can be implemented within existing systems without adding an external layer of automation like RPA. Sometimes emerging cognitive and artificial-intelligence solutions can deliver additional value. The key is to customize the solution design by exploring options beyond RPA. That means your team needs to know about automation tools and approach options. Successful automation efforts are more than twice as likely to deploy machine learning and other cognitive-based automation capabilities. The foundation for future success is a systematic approach to identifying sources of value and deploying automation technologies appropriately. (For more, read “The Automation Imperative.”)

Nearly 60 to 80 percent of data processing and data collection in the insurance sector, for example, have high technical feasibility for automation. At one company, the automation team considered the overall customer journey and how data were created, transferred, and ultimately used by each department. This required cross-functional collaboration with other departments, leading the automation team to develop a clearer picture of:

  • processes that needed considerable standardization or re-engineering prior to automation
  • processes that could be further automated using RPA
  • processes where the application of technologies other than RPA, such as chatbots and machine learning, were the right choice.

4. Manage the bot ‘brand’

RPA has the potential to unlock significant value quickly. Processes automated in a few weeks can result in tremendous improvement in process quality and reduce process costs by nearly 80 percent. But this can give rise to unrealistic stakeholder expectations about the value RPA can deliver. On the other hand, many other executives are still skeptical about using bots. Managing these very human reactions is key to making sure bots have a chance to succeed by having the discipline to implement them where they can be most effective. A poor understanding of goals, inadequate testing, or a misunderstanding of intent can lead to faulty process automation which can cause painful rework, poor resource reallocation, and inefficient bot maintenance, all of which impact the project’s overall ROI. Poor results mean stakeholders don’t sign off on their individual bot. Done often enough, that creates a brand-image problem for the automation team, causing fewer additional stakeholders within the organization to implement RPA.

 


 

Failing to effectively engage a digital workforce will result in an outdated employee skill set that will raise costs and make companies less competitive in the long run. The most effective route to turning around a flailing RPA initiative is to thoroughly evaluate the transformation objectives, approach, and team capabilities.

 

Marty Pavlik is a senior digital expert in McKinsey’s Chicago office, and Saurav Tripathy is a digital specialist in the New York office.