For decades, businesses have been looking for every opportunity to increase efficiency and productivity. Even as business evolves, employees spend far too much time on tasks that should be handled more efficiently. Consider
an IDC survey
that found data workers spend 90% of their work week (about 36 hours) on data-related activities such as searching, preparation and analytics. In doing so, they call upon more than six data sources, 40 million rows of data and seven different outputs on average. And they do this repeatedly, to answer new — and even previously asked — questions as they arise.
In today’s world, businesses can take advantage of automation to free employees from mundane, repetitive tasks such as these. Robotic Process Automation (RPA) is one of the technologies being applied in this way.
What is RPA?
RPA is software that assists workers by automating rules-based procedures and tasks. In other words, it is simple process automation often used in the form of bots that take care of some basic tasks on behalf of employees.
Some compare RPA to macros or scripts. However, while RPA is meant to mimic and automate mundane, repetitive tasks (much like macros), it can handle a high volume of complex processes. On the other hand, macros can only handle simplistic tasks. That said, like macros, RPAs can only carry out the steps they’re programmed to handle.
It’s also worth noting how RPA compares to
. Where RPA is focused on discrete tasks, hyperautomation is — as Gartner says — focused on using business-driven automation to make increasingly AI-driven decisions. Organizations can achieve this with hyperautomation by rapidly and systematically identifying, vetting and automating as many business processes as possible.
Back to RPA. Organizations must train their RPA tool prescriptively. In other words, if a task or process involves five steps, they must record them in that order. If a step changes in the process, the process must be re-recorded — i.e, the tool must be retrained.
As Forrester says
, “RPA, in its current state, does not handle variation well.” This is one of the reasons it takes a fair amount of time to deploy RPA in a complex environment.
In addition, each user must install RPA software on their computers, and all users working together must be on the same software version. For instance, in a finance department relying on RPA, every team member must have the same RPA software installed.
Once RPA records a process and then subsequently kicks it off on its own, it must spin up a virtual environment each time. It must then shut down the virtual environment once it performs all necessary functions. As a result, while an RPA might work incrementally faster than an employee, it’s a drain on enterprise systems and expensive to run.
Plus, an RPA cannot work dynamically. Employees must ask separate requests of it, which it processes separately. For example, “What is the latest sales forecast?” and “What are month-end bookings?” are separate requests.
Even RPA vendors are recognizing the importance of looking beyond simple task automation
to truly enable humans
The Benefits of RPA
Organizations take advantage of RPA bots to realize the following:
Higher productivity by efficiently taking care of core processes
Higher-quality operations due to fewer errors and more accuracy
Lower human error by ensuring tasks and processes are executed consistently
Improved employee experience by freeing people from tedious, repetitive tasks
Greater organizational agility by freeing employees to focus on value-add activities
Reduced costs associated with human error
Better compliance with regulations and standards that require accurate, consistent data
Better customer experience due to faster, more accurate processing of requests
Considerations Before Implementing RPA Technologies
Organizations interested in implementing RPA technologies should keep in mind these common barriers to success:
Choosing the right platform. Organizations must evaluate and select from a dizzying array of more than 45 products being marketed as RPA. Some of these fall into the Robotic Desktop Automation (RDA) category. RDA tools aren’t designed to scale because they must be deployed across desktops.
Lack of commitment from management. To deliver an impactful ROI, RPA projects must automate enough processes. Without needed support from executives and managers, these projects often get delayed or even shelved since the projects require significant IT involvement (such as to apply roles and permissions).
Trying to automate processes that are not fully mapped out and understood. Because RPA automates processes in a methodical, step-by-step fashion, it’s essential that everyone agrees on the right steps for the process to be automated. Otherwise, the organization might simply automate an inefficient process — which does nothing but automate inefficiencies.
Choosing a complex RPA tool. Not only will the organization need to train the tool, users will need to interact with it in their daily work. A tool that requires significant programming and is difficult to use often ends up seeing low adoption.
Automating a process that changes frequently. As explained earlier, an RPA tool must be retrained (or re-programmed) any time a process changes. Applying the tool to processes that change frequently can become a frustrating exercise that negates the efficiencies to be had by automating.
Failing to account for change management. People often need to change the way they work when RPA is introduced as they will no longer spend time on so many of the tasks that consumed them. The right stakeholders need to clearly communicate with and support these employees as their roles transition.
Not focusing on business outcomes. It’s not enough to automate for the sake of automation. The project needs to be clearly tied to processes that are essential to the business and that will deliver measurable benefit by being automated.
Common barriers to Automation Adoption
What to Look for in an RPA Solution and Vendor
With so much riding on the success of an RPA implementation, it’s essential to select the right solution and vendor. Make sure the solution and vendor satisfy these requirements at a minimum:
Automation of complex human workflows. Ideally you want to apply automation to some of your most sophisticated processes — the ones that deliver significant value to your business but require so much time of your employees.
Security focus. Make sure the solution only stores transactional data and erases it as soon as the process is complete. Many RPA solutions store sensitive data. Some even send sensitive data to third parties like Google Cloud, Microsoft Azure, and Amazon Web Services to process each request. Considering all the sensitive information that can be part of a process, this exposes your organization. Simply put, in this way, “
RPA introduces a new attack surface
that can be leveraged to disclose, steal, destroy or modify sensitive data and/or high-value information, access unauthorized applications and systems, and exploit vulnerabilities to gain further access to an organization.”
Cloud processing. Look for a solution that works completely in the cloud (while offering on-premise solutions if that’s your preference). Many RPAs are dependent on Virtual Machines and all users being on the same versions of software, surfacing all the challenges organizations face trying to coordinate complex, on-premise infrastructure.
Embedded Natural Language Processing (NLP)/Natural Language Understanding (NLU)/Natural Language Generation (NLG). Without the inherent ability to naturally process, understand and generate language as it is spoken naturally, RPA tools are dependent on third-party tools for Optical Character Recognition (OCR) and text extraction — among others — to execute their automation. Again, this introduces complexity and the potential for breaks in the tool when incompatibilities arise between the various third-party tools working in the background.
In-house Machine Learning algorithms. Similar to the point above, when a solution relies on third-party Machine Learning, it is more susceptible to issues and breaks.
Integration with messaging platforms. Look for a solution that provides a conversational engine so users feel like they’re talking to another human when interacting with the tool. Many RPA tools lack such a capability, making for a very unnatural user experience that discourages adoption.
Easy and quick to deploy. Be sure you can deploy the tool quickly. It should be possible to get started using it within just 30 minutes for a simple process. Even for the most complex process automation, you should be up and running within 10 business days. Beware of solutions that take weeks or even months to program and deploy as this can ultimately undermine the potential ROI.
Typically, RPA implementation projects involve three stages:
1. Assessing current state and determining what processes could be automated
Many organizations go wrong in step one because they don’t properly document current processes, and/or choose very complicated processes that can’t easily be automated. Much like the “garbage in, garbage out” principle, thinking carefully about “what goes into” automation has a direct impact on “what comes out.”
Unless adequate thought is given to what parts of the business are suitable for automation, RPA projects fail to produce anticipated returns. Plus, the business finds itself revisiting the best processes to automate, leading to incredibly lengthy projects. Strong candidates for automation includes processes that are:
Well-defined, based on rules with limited exceptions
Consistent; that is, the process is not changing very frequently
Not part of a system or technology that is planned to be replaced or upgraded in the near future
Not overly complex (ideally, the process calls upon data from 5 or fewer systems)
2. Implementation / pilot
Once you’ve selected the processes to automate, your team decides where to start and how the project will be run, often with the help of an implementation partner and/or the RPA vendor. The basic steps are:
Setting goals for the pilot. If possible, define quantifiable goals and targets that make sense for your project. Some common metrics to track include automation (# of competitions of given process, amount of data passed through) and accuracy (percentage of success completions without employee involvement / errors )
Pilot launch. Run the pilot as designed and carefully monitor results on a daily basis. That includes tracking key metrics mentioned above, and delivering regular reports to key stakeholders. Typical pilots require a minimum 2-4 weeks to run and verify the results.
Evaluate success of pilot. Once the pilot is complete, conduct a full postmortem consisting of a detailed report on all key metrics tracked along with qualitative feedback from internal analysts. When working with an implementation partner, use the data they provide, but verify with your internal team. If the pilot satisfies your success criteria, consider moving forward, or run additional pilots as needed.
3. Go live and ongoing maintenance
The process for going live will vary greatly for every business depending on the scope of the initial deployment. The key consideration is around ongoing management of RPA — the deployment is not the finish line, but rather the starting point. Successful adoption of RPA is a living process that requires consistent organizational support.
To ensure success, put in place a strong team to manage the RPA installation. A typical RPA management team (often referred to as the RPA Center of Excellence, or RPA CoE) consists of a business analyst, a solution architect, a developer, a project manager, and an executive sponsor to steer the overall program. The team has to stay ever vigilant to changes in business and market conditions and make necessary adjustments to both the process the RPA has automated and how RPA is being used to automate the process. They also have to consistently measure the effectiveness of the RPA solution and consider additional processes or additional parts of the same process that will maximize long-term ROI.
A Better Way to Leverage RPA?
A typical RPA implementation project going through the steps above can often take 3-6 months. A more modern approach uses
as an easier entry point into the world of automation. Unlike traditional RPA technologies, Digital Employees can be deployed by business users for specific use cases in functional areas like Human Resources, Accounting, IT Service Desk, and Sales Support without heavy planning and project management. Digital Employees link your human employees with the systems they already use (e.g., BambooHR, DocuSign, Jira, SAP, ServiceNow, Slack, Salesforce, and Zendesk), and are flexible to adapt to the needs of the business. Plus, because they can be rolled out and trained in 2-4 weeks. SKAEL customers often realize ROI on a single Digital Employee within 45-60 days.
Interested in seeing if SKAEL is a fit for your business?
Request a demo
The SKAEL Universal Adapter: Connecting our Digital Employees
Natural Language Processing: What It Is and Why It Matters
Robotic Process Automation – What is RPA?
What are Digital Employees?