You’ve done it a thousand times. Walking through operations, you’ve asked yourself “Why am I paying these smart people to spend half their time doing mindless tasks and data entry? What if I could free up their time to allow them to do what they do best – solve problems. That would give them more purpose – that would give our business more purpose”. Ok, maybe you didn’t think those words exactly, but I can guarantee that if you’re a high-performing company, it was something along those lines. Although detention and delay times are the items with the largest targets on their backs with respect to operational efficiency, you don’t have to look to far for the next one on the list – manual data monitoring, entry and management.
We generate endless volumes of data each day in our dispatch and accounting systems. Even more is found in unstructured formats, such as e-mails, and invoices. Many paid hours are spent taking information from one format and rekeying it into one or more other disparate systems. This duplicated effort is not only expensive, but it opens the (large) possibility of errors through typing mistakes or missed documentation.
The reality is, many of the tasks handled by our operations, accounting and customer service teams are repetitive and prime candidates for automation. In general, any repetitive, rule-based task is a prime candidate for automation. So why haven’t they already been automated? Until recently such automation required a lot of expensive programming that resulted in systems that were not very flexible. Simple upgrades of your dispatch system could cause the automation to fail simply by having a table renamed. When this happens, your team must revert to the old manual processes until the programmer has the updates in place. The other reason is human nature. The growing list of portals and manual updates downloaded to today’s bus companies is like compound interest. It started with a work-around to make one customer happy and has grown exponentially to the elephant in the room that no one wants to talk about.
Web portals have become more common. Documents such as invoices can be easily uploaded. Documentation supporting additional charges can be transmitted, and many can log comments when further explanation is required. However, they can introduce other inefficiencies. A human is still needed to transfer or transpose the information into the portal. Errors are still possible and very likely, despite best efforts. Depending on the internet speed where that employee is located, a lot of time could be wasted if your team member must wait for the information to get uploaded to the site and then wait again for the next screen to load.
The Institute for Robotic Process Automation defines RPA as “the application of technology that allows employees in a company to configure computer software or a ‘robot’ to capture and interpret existing applications for processing a transaction, manipulating data, triggering responses and communicating with other digital systems.” In more practical terms, it’s software that you can “teach” to move information from multiple inputs to multiple systems without the help of a human. RPA is designed to be deployed in days or weeks, not months. It also does not care where the information comes from or where it goes to. One of it’s strengths is that it does not rely on things like API’s to bridge into other systems or require complex programming to work.
Let’s look at a simple example that happens many times a day in your company. A customer e-mails a charter or field trip request to your CSR team. The CSR is constantly monitoring their e-mail for these requests. When they see one, they are likely saving (sometimes printing) the email to get the request on file. The CSR then creates the order and enters the charter details. Once it is in your dispatch system, the CSR then replies to the original e-mail to confirm that the order has been entered and received. Depending on how busy it is, each CSR may be doing many of these transactions every day. Each of these steps could be automated by RPA. The robot can monitor the email queues more efficiently than a human can. It will parse each e-mail and pull out the new charter request information. That data is then moved into your dispatch system where it checks availability and schedules the pickup date, time and location. Because RPA is rules based, if there is missing information or any other exception, it will flag the appropriate CSR to handle it. Once there are no exceptions, the system will generate and send the customer email confirmation. In this one scenario you can see that the CSR will now have time to work on higher value-added tasks, likely making their work much more interesting. The process will be handled more efficiently, and the possibility of errors is greatly reduced. The best part is you can implement entirely on your own, no need to get in the long queue with your dispatch system provider to complete it on your behalf.
So, what should you pay attention to when considering an RPA implementation? A recent CIO.com article gave 8 keys to having a successful RPA implementation.
- Do the Research – Frank Casale, founder of the Institute for Robotic Process Automation & Artificial Intelligence, recommends that you invest the time to build a business case for RPA and learn about the products available. There are three key boxes to get success and only having two out of three will not work. The boxes include:
- Choosing the right technology solution to meet your organization’s needs;
- Creating a solid business case for RPA, including developing ROI metrics;
- Assessing current processes and organizational issues to avoid political problems. Keep in mind that RPA is a disrupting technology. People will fear that the technology will affect their jobs, possibly eliminating them. The human resources that will be affected are not going to be eager to train the robots that might replace them. A vision and roadmap for the future needs to be communicated and it should include new opportunities for displaced workers. Managers also need to go in with open eyes, a cautiously optimistic mindset. Management of expectations is critical.
- Educate Staffers About RPA – It’s important to clarify what the technology will and will not do with regards to employees’ job roles. Many organizations use RPA strategically by helping staff do routine tasks quickly and efficiently so that they can spend time addressing higher priority needs. Determine how your company will use the technology and then honestly communicate what that vision is so that you get people onside with you before rolling the technology out.
- Determine Where the Technology Will Work Best – You will want to identify processes where you are most likely to see a positive business impact. Keep in mind that it will not always be easy. However, as the organization becomes more experienced with RPA, other processes will more easily be uncovered. Many successful implementations are very selective as to what processes to automate – look for something that it repetitive and frequent. Once you have achieved some initial success, fight the urge to try and automate everything. If a task can not be done without a very limited amount of human interaction (preferably none), then it really is not a suitable candidate. When picking your first process, remember that while cost reductions are important, improving customer experience is even more valuable. The client experience is what will be your competitive advantage.
- Keep It Simple and Modular – RPA works best when it is not complex. As much as possible, keep your bots as generic, common and reusable objects. Mona Kahn of Fannie Mae recommends that you externalize all variables and logic to minimize failure points. This makes it much easier to update when something changes and makes it much faster and easier to test before putting the changes into production.
- Don’t Neglect Data Security – Make certain that processes can not be manipulated. Start with any processes that are business or mission critical. These must be secured before any implementation is attempted. Regardless of how critical a specific task is, keep in mind that RPA will process much faster than a human can, so ensuring that a bot is secure must happen during the testing phase and this should be a show stopper if it is not secure.
- Test Implementations Regularly – Notwithstanding what we just discussed, testers can only try to test so many points of attack. Weaknesses are going to appear once the robot has gone live. Keep in mind that individual testers may not be as experienced with a process and may only be looking for positive results. If possible, include the staff that actually do the task in the testing phase. Additionally, test the automation on desktops running legacy systems to ensure that the desktop will be capable of handling the infrastructure requirements. It is critical that you understand how different bots work together so that processes do not break.
- Develop a Cross-Functional Center of Excellence – By this, I mean put together a team that will share experiences and best practices to other parts of the organization. The idea is to leverage previous efforts to ensure that the wheel does not get reinvented repeatedly. Make sure that both failures and successes are documented to make certain that the knowledge of those experiences is not lost.
- Prepare for Future Advances and Challenges – RPA is going to advance, and each company must keep up with the changes. Eventually, each organization will struggle with the management of the automation. End users will find ways to automate desktop processes for their own use, so understanding how the introduction, elimination or upgrading of enterprise applications will affect these deployed bots will become increasingly critical. Like any other asset, bots need to be tracked and managed so that they can properly be maintained. Consider what would happen if a password policy was changed for one of the applications the bot uses and how would you know if data is not being properly passed though?
So, what is driving the implementation of RPA? The obvious one is cost reductions. Forrester Research estimates that about 16% of US jobs could be replaced by RPA by 2025, while creating new jobs that are equivalent to 9%, meaning a net loss of 7% of jobs. This is because bots are generally low cost and relatively easy to implement. In the financial services sector, there is a major shift away from manual, clerical-type jobs and a move towards more analyst or advisory jobs. The shift is being fueled by the vast amounts of data that RPA is creating, resulting in customers needing more human advice to help them process it. In some cases, bots are assisting the human advisors by identifying patterns and offering options that the advisor can then present and explain to their clients.
There are several companies that have experience in working with our industry. These include:
- Kofax with it’s products that include Kapow, TotalAgility and Information Capture
- Pega with it’s Pega Infinity digital transformation suite
- Jacada with it’s Jacada Agent Desktop Automation
- Automation Anywhere has it’s Automation Anywhere Enterprise Suite
- UIPath – the UIPath Enterprise RPA Platform
- Nintex Platform – utilizing Promapp, DocGen, Xtensions Framework and Hawkeye applications
- IBM – Robotic Process Automation with Automation Anywhere
For those with an in-house development team (even a small one), building your own RPA toolkit and ‘army of bots’ is possible. There is even an open-source development framework to get you started. This Python-based framework – Selenium, is used by thousands of companies all over the world. Its worth examining if RPA is on your to-do list for 2019.
Any of the companies or open-source packages that offer RPA tools, can allow your company to achieve similar or even greater success. The key is to know what it is that you want to accomplish – without a defined project scope there is a significantly reduced likelihood of goal achievement. In fact, without spelling out what will constitute success and how to measure it you are probably dooming the project to failure right at the start. It also means that you are more likely to use someone or something other than the optimal solution provider. Similarly, the management of expectations is another critical component. Your people need to understand that it will take some time to optimize and implement these bots. If someone is offering to come in and “have you up and running in a week”, be very skeptical and ask some very pointed questions as to how that will happen. It is possible that they have extensive industry specific experience and have semi-customizable templates that can get you implemented quickly. If they tell you that “customer service is customer service” then they probably are not going to work. A quick example is with scheduling charters. If an RPA provider does not understand how HOS can impact transit times, then they may program a solution that ignores the regulations. A lack of industry experience does not have to be a deal breaker, but it will mean that your project team will have to spend more time and be more explicit in their requirements and specifications document.
So, in summary, some quick takeaways on RPA are:
- Robotic Process Automation is a potential game changer for our industry, possibly causing a revolution in how we handle routine customer service tasks.
- People are going to be afraid of losing their jobs when they hear the terms “robot” or “automation”. Expect that reaction and have education and communication resources prepared to help people get over those fears.
- The potential for head count reduction is there, but it should not necessarily be the driving force behind this initiative. It is more likely that you will be able to handle more productivity out of the same number of people.
- Do not treat automation as an ad hoc process. Have some form of control over where bots are implemented and try to use common programming as much as possible. This will allow you to know what needs to be updated when other systems are upgraded. Additionally, it will greatly reduce the time needed to develop fixes when problems arise.
- Start with processes that occur regularly so that you have a few highly visible wins to the start of the project.
- Expect there will be setbacks. Many of these processes have dependencies and it only takes one to cause a failure. Involve the people who currently do the job in the testing phase. They will have seen the process breakdown before and can offer insights as to where to challenges may come from. This in turn will allow for more robust testing pre-implementation. It will also increase the acceptance of RPA as they will have some say in how it is implemented.
- RPA is not a magic bullet. Managing expectations is critical. Exceptions are going to happen. Some processes will consider having 40 to 50% of appointments scheduled by the bot a success. Other processes may achieve closer to 95%. Some processes may require human interactions. Your team needs to be prepared for these to happen. At the same time, let them know what that 40% means in terms of freeing up time to handle their other tasks that might be getting pushed off to the side.