メインコンテンツに飛ぶ

We'd prefer it if you saw us at our best.

Pega.com is not optimized for Internet Explorer. For the optimal experience, please use:

Close Deprecation Notice
rpa issues

The 3 critical mistakes of RPA (and how to avoid them)

Nolan Greene,
共有 Share via X LinkedInで共有 コピー中です...
ログインしてブログを購読する

The hype machine around Robotic Process Automation (RPA) has been working overtime. With promises of fast deployment and empowerment of business line workers to eliminate the rote task work that plagues their day-to-day operations, it is no wonder that many organizations worldwide are exploring and trying to implement or scale RPA. But listen more closely and you’ll hear whispers about what’s not going well with RPA – slow deployment cycles, broken bots, and deltas between beginning goals and end results.

Don’t get us wrong – RPA is a fantastic technology that bridges the gap between legacy applications and business process optimization where there are no APIs – and we see enterprises achieve tangible, positive results with RPA every day. That’s because these enterprises have avoided the most common mistakes of today’s RPA deployments.

The three critical mistakes of RPA

1. Believing that RPA is easy.

This is perhaps the most common myth around RPA – enterprises expect to purchase a solution, train business users to build automations, and see the ROI start to accumulate. While there are certainly results to be had from training business users to build automations, the truth of the matter is that building sustainable automations that fit into broader business processes and systems is more complicated. In fact, in a recent Pega survey of enterprises deploying RPA, 50% of respondents reported that RPA was harder to deploy than initially expected.

For most high-impact processes there are many human variations that must be accounted for. Many of these are undocumented or unknown, and it takes a deep comprehension of how work gets done to truly understand and plan for these. Application variations are another factor to be considered; from custom UIs to third-party applications out of the enterprise’s control, these variations can create considerable complexity.

Yes, it is possible to train business users to build simple automations. However, these automations only go so far when it comes to getting measurable results. If they operate in isolation and don’t fit into broader departmental or organizational processes, they likely will not provide any real impact. Even worse, they are very prone to break as underlying business systems, rules, variations and processes change. Gartner makes this point in their July 2019 Magic Quadrant for Robotic Process Automation Software, highlighting the technical debt that enterprises face when a third-party system within an automation upgrades or changes. A change to the third-party app or system can have serious effects on even simple automations.

Finally, don’t forget about governance, security, compliance, or hardware infrastructure needs. These are critical to RPA success, and the way they are implemented needs to be consistent across operations.

Bottom line: Enterprise-scale RPA isn’t as easy as some would have you believe, which is why business leaders really need to understand the outcomes to be achieved, use-cases to be prioritized, and stakeholders and systems to be affected. Business and IT collaboration is essential in helping to understand these variables and identify the right RPA opportunities.

2. Not taking an agile approach

Traditional unattended RPA is generally not very agile. There are perhaps a few simple tasks that can be automated in an agile manner this way, but for quick wins, task automation, attended RPA (think bots that assist workers on the desktop in real time), and RPA unattended, combined, is the most agile way to implement RPA and at scale.

Consider the desktop of your call center agents and other back-office transactional workers – these workstations are often replete with legacy applications geared toward completing multiple, similar, manual tasks that are necessary for completing processes. Given the number of employees doing this type of work in most organizations, and the quantity of the applications and tasks performed, applying attended RPA to these delivers better customer outcomes and generates business value quickly.

Pega has found the success of this agile approach with attended RPA provides a springboard for continuously and iteratively building and scaling an RPA program. In our experience, initial projects are built and go live within 6 to12 weeks. Each iteration of the next set of task automations are typically then delivered in 1 to 2-week cycles after. In one case, a client scaled up to 35,000 attended RPA bots in the first year.

One caution – not every RPA vendor promising attended capabilities delivers the real and non-invasive, assisted automation technology. “Watched RPA unattended,” where unattended RPA technology is deployed on an employee desktop but doesn’t allow for the user to touch the mouse or keyboard, is not RPA attended and does not, therefore, deliver the type of value at scale as true desktop RPA. (But this is a topic for another blog!)

3. Treating RPA as a platform

RPA is not the endgame of intelligent automation, rather it is one piece of the puzzle. Many fall for the myth that RPA is a fully-baked intelligent automation platform but the truth is, you cannot build transformation on a “platform of band aids.” Enterprise systems are complex. Automating and streamlining the complexity of multiple, long-running processes that interlink internal and external systems, machine and human work, bespoke software, and third-party apps and systems is not what RPA was created to do. That is the work of an intelligent business process management software (iBPMS) platform.

Nor is RPA a platform for artificial intelligence. While RPA has been touted as a “form of AI,” it quite literally is not – RPA simply automates legacy UI where an API doesn’t exist. What RPA is adept at doing is automating work that meets all three of these conditions: high-volume, low-complexity, and rules-based. Business leaders typically discover there is less work within the enterprise that meets all three criteria than they thought.

How to succeed with RPA

Whether you’re new to RPA or figuring out how to move forward on an existing project, avoiding the mistakes above will help ensure a positive trajectory for your RPA projects. To further align your RPA project for success, we recommend the following:

  • Don’t get hung up on “easy.” There is a place for business users building automations, but to get to real scale and transformation beyond the limited simple stuff, IT needs to work together with the lines of business to design an end-to-end intelligent automation solution that ties in to pre-existing business systems and accounts for their sophistication/complexity.

  • Look to the desktop for quick wins. If generating ROI at scale and quickly with automation is the objective, using attended RPA on the desktop will almost definitely help your organization. Agent desktops are often consumed by API-less legacy applications that don’t talk to one another. RPA bridges the gaps between these applications in mission-critical processes. This improves employee experience and customer outcomes, while generally demonstrating measurable results in just a few months.

  • Leverage an intelligent automation platform. RPA can achieve a lot but it needs the power of an intelligent automation platform to reach its true potential. Low-code process design, strategic AI integrations (remember: RPA is not AI!), and the ability to augment processes with document automation bots, email bots, and other channel tools benefit from the end-to-end orchestration and scalability that only a unified iBPMS platform provides. This unified approach to intelligent automation also helps prevent underlying systems changes from breaking your bots.

The statistics don’t lie

As EY reported in their paper, “Get ready for robots,” they’ve seen as many as 30% to 50% of initial RPA projects fail. In our experience, we’ve found that few enterprises get automation to scale and many regret wasting 1 to 2 years, or even more, attempting to realize the big ROIs promised by RPA vendors. Pega’s survey found that only 39% of RPA bots are deployed on schedule.

Don’t be one of those enterprises that invests big money in RPA only to see low-scale results, or worse, wind up abandoning an RPA project altogether. RPA success is attainable. Remember, RPA is standalone technology that excels at automating high-volume, low-complexity, rules-based tasks. RPA tools work best when they complement more broadly-based automation tools. Digital process automation (DPA) and iBPMS platforms are designed to orchestrate and streamline complex enterprise operations, including RPA deployments.

Learn more on how to succeed with RPA:

タグ

トピック: ロボティックプロセスオートメーション
製品エリア: ロボティックプロセスオートメーション
課題: エンタープライズモダナイゼーション

著者について

Nolan Greene is an industry analyst and marketer focused on digital transformation.

シェアする Share via X LinkedInで共有 コピー中です...
シェアする Share via X LinkedInで共有 コピー中です...