Automation is key to workflow, and it’s hard to have good automation without some innovation. See what our executive panel had to say about the way automation is changing the business environment, whether robots are really taking over our jobs, and more.
We’ve noted a trend in automation towards ensuring continuous improvement – what measures can you take to build this into your automation efforts?
Sam Babic: You cannot understand if you’re improving if you don’t understand what you’re improving toward, so building continuous improvement starts with setting a goal or outcome for your automation. The goal should be as SMART as possible — specific, measurable, achievable, relevant and time-based. “Reduce manual entry into third-party system by 60% in the next quarter” is a good example of a SMART goal. Something like 100% reduction in manual entry may not be achievable, and simply stating a 60% reduction without putting a timeline around it leaves the goal open-ended. Keeping goals measurable is key to tracking, so ensuring that the automation system has the ability to log and generate reports is critical to achieve continuous improvement of an automation goal. From there, you can monitor the process continuously and conduct checkpoints where you can tune the process over time. In some cases, the system may be intelligent enough to tune itself, but should be transparent enough to indicate such tuning.
Chris Huff: One measure that most overlook, but that is absolutely critical to establishing a repeatable and scalable automation program, is to build a Digital Management Office (DMO) (aka Center of Excellence) that includes continuous improvement among its list of responsibilities. Additional responsibilities include creating strategy and governance, effecting change management, monitoring performance management and reporting, standardizing tools and technology, and pushing innovation. The DMO should work with other established offices, such as the Business Management Office (BMO), which is typically responsible for finding enterprise inefficiencies and enacting optimization plans to drive productivity. These offices typically have process experts that largely leverage Lean Six Sigma (LSS) to continually optimize operations. This is a great office to include in the larger digital transformation effort that’s typically driven by a DMO. Bringing together the BMO and the DMO can be a powerful combination that unifies complementary methodologies and tools to accelerate enterprise value creation.
Bruce Orcutt: Automation is a task-driven imperative with a sole purpose: to get machines to do the repetitive tasks we creative, free-thinking humans don’t want or really need to do. The job isn’t over just because your digital workers have been deployed. It is necessary to monitor and track performance of the process post-implementation. It is absolutely critical to enable continuous improvement by monitoring and measuring automation’s up and down stream impact to ensure ongoing protocol compliance and preventing the “bottleneck shift” and preventing the possibility of negatively impacting the process in other places. Monitoring the digital workforce as well as the entire end-to-end process post-implementation is just as essential as the planning and execution. As reported by Ernst & Young, it is this stagnation post-implementation that contributes to the failure in many RPA initiatives.
Once automation has been implemented, it is likely that not all of the bots will be in use on a day to day basis or across all processes. It may be that the bot is only triggered in a very specific instance. Automation will continue to act based on the set of rules that they were coded for. The digital worker has no concept of whether it is creating good or bad outcomes. Therefore, it is important to monitor the overall performance and ensure positive outcomes. And no, that doesn’t mean that someone needs to watch a dashboard waiting to respond to a bad outcome. Unless there is already a sophisticated ability to plan, track, and monitor process behaviors (doubtful) there will be a need for technology to automatically detect specific conditions and alert to issues in real time. If not, the reliance on manual efforts to review and monitor execution post-implementation will be quite clear; and what type of digitalization and automation is that?
When is it time to bring in AI in your process automation efforts?
Huff: Once the problem you’re trying to solve is defined, I don’t believe it’s ever too early to bring narrow AI or “fit for purpose” AI into the effort. I also believe it’s pragmatic in the age of waterfall and agile to execute optimization and automation initiatives concurrently. An example is applying AI-powered cognitive capture Bayesian classifiers and neural networks to ingest unstructured data, such as emails, PDF images, and documents, to determine data relationships and perform the transformation to structured and standardized formats. Once the AI-powered technology has accomplished the heavy lifting, we can hand the structured data to simple rules-based automation solutions such as Robotic Process Automation (RPA). When deploying optimization and automation concurrently in this manner you tend to quickly expose how much variation and inefficiency exists within a given process, which will lead you to the next iteration of AI/automation solutions required to drive even greater straight-through automated processing.
Orcutt: We focus on applying AI technology to solve specific customer problems when processing unstructured content, specifically with content IQ skills that have the ability to digitize, classify, extract and enrich the data with understanding and meaning so enterprises can turn unstructured information into structured, actionable data. A digital transformation requires a plan on where AI fits in, but it is important to apply a focused approach and consider how AI technology can complement broader automation solutions both in the back office and customer experience. Furthermore, when starting an automation initiative there are several questions that come with starting the project. What process should be a candidate? What will its value be? How much time will we really be saving? Answering these questions takes a complete understanding of how the business processes truly operate and why, and is also a place to bring AI into automation efforts.
Talk about an automation attempt that went awry, and what was learned.
Babic: Automation increases efficiencies of repetitive tasks. However, if the automation is not configured to deal well with exception cases, you can run into situations that are difficult or tedious to unwind. When pulling results from one system and posting the information to a secondary system the phrase “garbage in, garbage out” comes to mind. So if there is unclean input data, or if there are exceptions to one step in the process and those exceptions aren’t handled appropriately, they can cause issues later in the automation. For example, when reading from each row of a spreadsheet and then posting the data to a secondary system, but there is bad data in the spreadsheet, the automation needs to recognize that. Otherwise, you are potentially cleaning up hundreds, if not thousands, of bad entries in the target system.
Huff: Success is built upon the backs of many failures … and we had plenty of challenges in the early days of automation. One that stands out, during my early days at a leading global consultancy, was a very large public sector customer who wanted to implement RPA because they’d attended a conference and heard how easy it was to implement. They heard grand stories of quick deployment, rapid and outsized return on investments, and that scalability was only inhibited by their imagination. My team was asked to respond to an RFP, and in the response I made sure to indicate that my teams only implemented RPA alongside the creation of a sustainment office called the Digital Management Office. The DMO is intended to make sure the AI/automation program is not a one-trick proof-of-concept pony, but rather a central piece of a larger organizational digital transformation strategy and enterprise capability.
Needless to say, my customer wasn’t excited that my proposal was a little more expansive, required a few additional weeks of work and in the end was slightly more expensive. While the board didn’t award my team the work due to a lowest price technically acceptable award methodology, I did maintain my trusted advisor relationship with my customer. Fast forward one year. The customer had challenges scaling beyond a very controlled-environment RPA proof of concept due to environmental complexities, so my team was called in to recover the program. This happens more times than most think. My advice is to be prepared with a final desired end-state and if it involves a scaled AI/automation program, then you’ll need to look beyond simple RPA while creating a governance and sustainment mechanism such as a DMO.
How can the SMB be a part of the automation evolution?
Babic: Automation doesn’t have to be complicated. Even SMB organizations can opt into automation by leveraging simple tools built for the cloud. Simple tools like IFTTT, Zapier and others can connect an SMB, its devices and its customers in an easy and elegant way. So for customers that don’t require the power of an enterprise-grade automation solution, it can be an easy way to participate in the automation evolution.
Steven Burger: Often, it seems bleeding-edge advancements are all for the enterprise, with benefits that depend on scale, large upfront investment, or both. The future of work is changing for everybody, not just the largest enterprises. Small and medium-sized businesses need the tools to keep up, in terms of driving productivity and efficiency, as well as in terms of providing the kind of digital workplace that attracts today’s workers. One way we’re seeing that addressed is in the form of cloud subscriptions and as-a-service models. These provide affordable, scalable versions of enterprise-grade tools, expertly implemented and managed by vendor partners. A good partner understands all of the different technologies, needs and obstacles at play in their customers’ environments and helps develop, implement and manage holistic solutions that take all of that into account for tailored, continuously optimized solutions. SMBs deserve a seat at the table, and these kinds of services help get them there.
Huff: Cloud-enabled solutions are enabling small- and medium-sized businesses that can’t afford the infrastructure and compute needed to automate across various regions, and with higher levels of complexity. SMBs are also benefiting from new consumption-based pricing models that allow a shift from capital expense (CAPEX) to operating expense (OPEX). Couple this with the mounds of experience that larger enterprises have gained and shared with the market, and SMBs now have a clear path on what to do while embarking on their automation journeys. Additionally, global consultancies and system integrators have a lot more experience than they did five years ago, so they have accelerators, templates and best-in-class frameworks that can help mitigate risk, lower total cost of ownership and accelerate time to value.
What are some of the coolest tools for automation that you are seeing out in the market right now?
Burger: There’s a use case we’ve been deploying in Japan that I’m very excited about. In the office, we’ve focused a lot on boosting computing power at the device, so different operations – OCR scanning, sending files directly into automated workflows or hot folders, securely printing from mobile devices, and so on – can be done quickly and easily at the device. We’ve combined that with optical technology to collect and process data from roads and upload it to a centralized, cloud-hosted map that lets municipalities check-in in real-time. We collect data from our optic technology at 30-40 MPH, analyze the big data, and make recommendations to city municipalities. If a road starts to exhibit signs of wear and tear, or experiences some sort of urgent issue, the city can see it coming and move quickly and intelligently to address it. It also helps intelligently triage roadwork.
Huff: Automation companies are stepping up their game by creating strong ecosystems of adjacent complementary technologies that collectively benefit the customer. One example is bringing forward cognitive capture to ingest data from documents, web and mobile, and transforming the unstructured data into structured format so that automation solutions like RPA can perform simple rules-based automation. Secondly, process orchestration is becoming a mainstream capability in all automation programs and handles the more complex non-rules based case management, exception handling and orchestrating the work between automated solutions and human employees. Coupling cognitive capture, RPA and process orchestration is proving to be a powerful combination to drive significant value across a broad range of use cases, while also setting the stage for the application of AI-powered and advanced analytics to drive even greater value.
When automation is introduced to a process, where are bottlenecks most likely to occur?
Babic: It is wonderful and rare when you can automate 100 percent of a process. Typically, within a workflow there is some necessary manual step. In cases where automation is not 100 percent, that bottleneck is often in the manual process. This sounds obvious, but if you go through the automation effort only to find that the majority of the work has to occur in a manual step, then you’re in for a surprise. In manual steps, people don’t often consider load balancing — meaning, if there is a manual step, what happens if that person is on vacation, leaves sick for the day, or simply takes an extended lunch? Understanding the manual touchpoints in the flow is key to understanding bottlenecks.
Burger: Automation can work wonders. But too often, when automating a process or workflow for the first time, people think about how to get from A to B without considering how we got to A or what happens after B. Even if the right software and implementation were used for automating a process, it’s vital to look at the ecosystem surrounding that process. This is another area where partnering with experts really shows its value. They’ve not only considered but seen the edge cases that can derail your automated processes, and they have the fix.
Huff: Bottlenecks are surprisingly found where people remain in the process! This sounds odd, but in the dozens of large-scale implementations I’ve been fortunate to lead and advise on, this is predominately the case. Think about it. People are meant to focus on higher value areas requiring decisions and actions and/or adjudication. These areas of work simply take more time than a transactional task that requires no thought or judgment. Therefore, when automation is applied to transactional tasks, the cycle time is reduced while throughput increases. This means that any exceptions that are routed to people are also identified quicker. This results in a backlog of higher value work that requires the attention of an employee(s). To address this outcome we normally refine the rule set used to identify exceptions and minimize them. Additionally, organizations will typically use this time to deliberately seek to improve the customer experience, so they recruit more people to this higher value work that often involves interacting with a customer to resolve an issue.
Should people be afraid of losing their jobs to robots?
Babic: The purpose of automation is to automate tasks, not to automate jobs. There are many tasks that are repetitive and tedious, and honestly, some of those repetitive and tedious tasks didn’t exist prior to the introduction of technology. When everything was manual there was no need to copy information from one system and post it into another system. Technology increased the scale of data and data operations beyond what manual processes could have accommodated, thereby introducing repetition — especially where there isn’t an integration between systems. Doing steps manually is tedious, error prone and most likely not fulfilling to employees. Automating those types of tasks gives people the ability to focus on higher value endeavors. Automation can then give opportunities for employees to take their learning, and careers, to the next level. As domain experts of the tasks involved, those individuals are at the center of understanding what is tedious, error prone, and results in rework.
Huff: Society has been advancing since day one. Technology has become just another productivity tool used to empower people while adding capacity to organizations. People often don’t receive enough credit for the ingenuity required to constantly reinvent themselves with new skills, jobs and roles as a result of advancing technologies. From the bank tellers to the manufacturing assembly line, businesses have always created higher value roles for humans and worked around the technology. Some will say advanced technologies are edging people out, but I have faith that we’ll continue along the path that we have always journeyed. That path may take new turns, but we’ll continue to navigate as we have always done. Recent research from the World Economic Forum states that by the year 2022, 133 million new jobs will be created, while 75 million jobs will be impacted – leaving us with 58 million net new jobs. This is just another validation point that reinforces the creativity and entrepreneurial spirit of humankind, and we’ll continue to shift humans toward higher value, more purposeful work.
Orcutt: When you say “automation” people may hear “job loss,” but RPA isn’t new and automation has been happening for years. While it is understandable that if you tell someone that you’re going to automate a lot of the tasks they perform at work, they think that means they’ll be out of a job tomorrow, or soon thereafter. Now it is certainly true that a new wave has evolved, driven by automation, artificial intelligence and machine learning, and is emerging across the globe, but workers shouldn’t fear. RPA today is ultimately about using software to automate repetitive, high-volume process tasks that would have once required a human effort. RPA shifts the worker to more intellectual work. Instead of people spending hours completing manual repetitive tasks, they can leverage RPA to focus on work that requires human ingenuity and creativity.