part0009

Screen scraping emerged in the early 1990s to provide organizations with the ability to collect screen display data from one system and then use it to process or display the information on another system. For example, many incumbent banks used legacy mainframe systems for key business operations such as banking transactions and customer payments—and still do today. Given that these mainframe systems could not support new software technologies directly, screen scraping provided a viable mechanism to extract this data for processing by more sophisticated software tools and generate value-added data.

Screen scraping software was also found to be valuable for testing software, repeatedly simulating a specific sequence of keystroke steps based on pre-defined IT test scenarios. Rather than tying up a human tester to click the mouse button on a specific web page icon 1,000 times under various test conditions, a software robot could be configured to execute the same activity significantly more efficiently, recording the results to a separate output file. These early testing “robots” could be run throughout the day or even overnight to help identify quality issues in the software.

Workflow automation dates back to the industrial age and the emergence of manufacturing and mechanization. The information age provided an opportunity to leverage the emerging power of computing systems and their growing processing capabilities to transform a particular data input into an output, through a prescribed sequence of tasks. For example, a customer registers online for a new bank account, which in turn initiates a specific sequence of background steps, from checking the customer’s credit history and address information to sending an email to validate their details. The emergence of computing provided the capability to automate a growing number of steps in a workflow. Using these emerging software tools, creative IT back-office teams in sectors such as banking would spend evenings cobbling together solutions to automate significant transactional activities, such as creating thousands of bank accounts on backend systems overnight or automatically sending an email if one of the key systems they were supposed to be monitoring started becoming unresponsive.

As these early software automations started to eat their way into more critical business processes, they started to catch the attention of their managers. Running these automations on spare desktop machines that no one wanted, located under a developer’s desk, was no longer an option. These DIY solutions expanded beyond IT to find wider applications in functions such as finance operations, where there were plenty of high-volume, low-complexity activities that could be applied to the maturing automation software tool. Blue Prism, one of the early RPA software product companies, coined the term RPA, Robotic Process Automation, in 2012. It aimed to capture both the futuristic vision of robots beyond industrial physical machines and provide an alignment to their initial purpose of automating processes. In 2015, the IEEE, one of the world’s largest technical professional organizations, created the P2755 working group. Comprised of industry leaders, the group’s task was to agree standards and formally define terms relating to this technology. The term Robotic Process Automation was formally recognized, and since then terms such as bots and digital workers have also been commonly used. The addition of enterprise-grade features to these early solutions has been critical in driving significant adoption over the last few years, with RPA now being a widely recognized software technology term. Gone are the days of slightly dubious automated programs running on desktop machines that leaned precariously on the IT helpdesk operator’s windowsill. These solutions had been promoted to enterprise-class, hosted in corporate data centers and supporting sufficient levels of security, access, and availability to function across an organization. These solutions bring with them key capabilities to support the enterprise to automate work.

RPA has been adopted as a broader term to represent what are, in fact, two different types of operation: assisted automation and unassisted automation. Unassisted automation is the automation of typically back-office functions that are low-complexity, high-volume transaction processes. For example, RPA digital workers are commonly used in the financial services sector to support activities such as credit card approvals. Most of the business processes following a customer completing an online application are handled by a combination of different IT systems, both internal and third party, such as credit reference agency checks, and data files. The digital worker requires no human input or “assistance” to complete its prescribed set of tasks, from checking the customer already exists in the business to assignment of their credit card limit.

In many scenarios, the activities can be automated with only a small percentage requiring human action, triggered by a specific set of defined rules. For example, a customer might request a credit limit that has a higher level of risk associated. In this circumstance, the organization could decide whether or not the digital worker needs to send the customer an automated decline—not a great response for customer satisfaction. It could instead send a suggested new credit limit, or the digital worker could assign the case to a human adviser to discuss further with the customer.

In contrast, assisted automations are dependent on some level of human interaction, with the digital worker software typically located “on” the employee’s desktop as opposed to a data center. Typical applications for assisted automations are sometimes referred to as “digital assistants” and exist in the organization’s front office, such as contact centers where the improvement of operational efficiencies, alongside a significant growth in customer channels, continues to provide organizations with a number of challenges.

Given the plethora of different IT systems, a customer calling into the contact center who raises a relatively low-complexity query, such as a change to their billing payments profile or an update to their contact details, may require the agent to spend significant time interfacing with systems instead of really listening to customers. The ubiquitous tapping on keyboards and apologies as “the system is slow this morning” are all examples that we are far too familiar with. A desktop automation can be started which, at the mouse-clicked command of its call center master, is initiated to log into the customer’s record and pull up the last payment made or the list of products purchased. A list of the most common automated activities can be accessed by the adviser on their desktop screen, informed by analysis of the most common queries raised in the contact center, and selected to initiate action by the digital worker assistant.

A company within the home computing manufacturing sector was able to reduce the average handling time of calls (sometimes referred to as AHT) in its contact center operations function by over 60%. A large volume of the calls were from retailers raising common queries on faults, requiring advisers to follow a set routine of navigating a number of different systems for information to help resolve them. Once prompted by an adviser, the digital worker took ownership of the high-labor components of the business process, logging on to several systems, checking for existing queries on product reference numbers, and presenting the helpdesk adviser with a consistent set of information to support them in providing the advice.

RPA automation tools have been able to adopt a similar approach and, as such, gain traction directly from groups that typically have been more wary of IT solutions, such as operations and finance function teams. Many of the market-leading RPA solutions provide the capability to quickly build automations by using a visual studio editor, allowing components to be dragged and dropped onto a screen and the automation process to be stitched together. Many of the solutions also provide a desktop recorder capability. This works by capturing object properties, including their values—for example, completing a name field on a form or clicking a button, which negates the need for any coding.

These low- and no-code recorder capabilities within RPA solutions have been invaluable in providing much greater accessibility to resources internally and externally to build automations. But organizations have found very quickly that as you move beyond the simplest of automations, knowledge of key development principles such as coding standards, error handling, and interfacing with other systems becomes invaluable in being able to build robust and re-usable digital workers.

One of the key advantages of RPA solutions is the reduced time taken, and hence the cost, to build a digital worker as compared to traditional IT projects. But in many ways, that’s also RPA’s Achilles’ heel. Lured by the ease of development cycles measured in weeks, it has been all too easy for organizations, and in some cases their partners, to focus on the quantity of automations at the expense of quality.

One of the key benefits of building an enterprise-wide automation capability is the ability to reuse components and hence further reduce build and configuration cycle times. However, this reuse places specific demands on the initial design quality of RPA solution elements, which at the outset should take longer to develop. The main building blocks of a solution, typically created within a desktop development studio environment, include the components, objects, and processes.

The components represent the individual screen interactions, such as adding a name into an online form. The objects allow the solution to interface to other systems, such as logging on to an enterprise SAP system (a commonly used Enterprise Resource Management system used by large organizations). As highlighted earlier, if defined tightly and built correctly, you should then be able to reuse the “log onto SAP” object for other automations that require that same step in the automation. The process in an RPA build defines the flow of the automation, typically representing the steps the human may take, along with supporting business rules, data, and calculations. The abstraction layers of component, object, and process need to be combined in a way that follows clearly defined design standards in order to maintain high build qualities within an automation.

Exception management, or error handling, is a particularly key consideration when building RPA solutions. Just as the performance of an employee can be improved if there is information on the types of mistakes and errors made, the digital worker provides a valuable stream of performance data that allows the reasons for errors to be identified and therefore it informs a process of continuous improvement on the effectiveness and efficiency of the digital worker.

System exceptions are raised by the digital worker if a technical issue occurs that prevents it from completing its prescribed set of activities. For example, an automation may be halfway through inputting information into an account-receivable screen within an SAP application and suddenly the SAP system becomes unresponsive. This, of course, may be experienced by employees as well, since the digital worker connects to the IT system in the same way.

The automation may be configured to raise an error code identification number of, say, 427 whenever the SAP system is unresponsive. Once the digital worker is deployed, it immediately provides a data-driven reason as to why its work was unable to be completed. For example, the automation may have been unable to work due to system issues over the past week. The data might show that 90% of the system-exception issues raised were due to error code 427, which is because of the SAP system.

Business exception errors occur when pre-defined business rules are broken. For example, an automation may be configured to monitor a specific email address. On receiving an email with an invoice attachment, it could add the details of that invoice, such as the supplier, invoice number, and amounts, into the company’s financial system. The digital worker may be configured to raise a business exception error-identifier code 873 if the total invoice value of the data being extracted is greater than 5% of the previous invoice from the same supplier.

This data can be ingested into the company’s target data repository, such as a data warehouse, providing valuable insights into the business behavior of a process. A large percentage of business exception codes 873 “Supplier total invoice value breach” errors could indicate unexpected variances in what should be relatively stable third-party costs or a cluster of issues with specific suppliers.

Despite the numerous resource management and activity-logging systems organizations have built over the last decade to help understand and improve employee performance, they will never come close to the value of data and insight that can be generated from digital workers. This provides a unique opportunity not only to continue to improve operations but to provide a valuable view of the general performance and maturity of IT systems. If office chatter consistently highlights the frustration of employees not being able to work because the systems “were down” or not working correctly, the digital workers will be very specific in providing a non-political, data-driven perspective.

Given the enterprise may have a wide selection of automations operational at any point in time, working on different business processes across different business verticals, the ability to monitor and control automations becomes a business-critical function. Typical features of RPA command centers, sometimes called control rooms, include business support, scheduling, access management, exception monitoring and management, audit management, and reporting—both historical and real-time. The automation control room needs to be integrated into the organization’s existing IT service management procedures to provide internal business customers across the enterprise with seamless levels of service.

Automation software can help accelerate an organization in its path to digital transformation, from supporting the rapid backlog of digitization initiatives to enabling wider transformation. RPA’s low technical complexity, speed of deployment, low cost, and tangible benefits have all supported its increased adoption across all major industry sectors. The innate ability of RPA to be flexibly implemented across the organization, be integrated with nearly all the IT systems, and solve a variety of business challenges has led to a wide range of use cases.

Organizations have predominantly targeted automating business processes that have high associated service costs and which are simple, repeatable, and predictable. These business processes typically either have large assigned employee numbers or have been the sweet spot of BPOs (Business Process Outsourcing). Automation provides the opportunity to reduce the time spent by employees on these activities or to directly reduce third-party costs. Automation can be applied to a vast array of rules-based processes across a wide range of industries and across different functions within the enterprise.

Manual data copying is sometimes called a “swivel chair” activity. The business processes typically involved require the desktop user to shift between two or more IT applications to update them, copying and pasting the data items. Swivel-chair activities have been the poster child for RPA, given their applicability to so many broken processes across the organization. Processes that involve repeated data entry are exposed to human error rates and result in lower efficiencies, increased costs, and constraints on the effectiveness of supporting IT systems.

For example, an organization registering a new customer via its helpdesk may require advisers to enter customer information into the customer database (such as a CRM system) and a separate billing system. There may be numerous reasons why the two systems are not connected via a traditional systems interface, such as a recent merger between two companies or technical limitations. But the impact of an employee incorrectly entering information extends beyond process inefficiencies to potentially incorrect customer billing.

For example, prior to an insurance claim being assessed by a claims adjuster, there are a number of administrative activities required in the early stages of the insurance process. A digital worker applied to a claims workflow may be set tasks such as validating that the customer has an active policy if the claim has been submitted online, and sending an email or text update to the customer once the policy claim assessment is underway. It could even contain a set of rules to automate the assessment of low-value, simple claims. The digital worker facilitates communication with all interested parties, securing the necessary approvals while improving the fluidity of the workflow by reducing bottlenecks.

As the quantity of data continues to grow, so does the importance and challenge of making use of it with the provision of accurate, timely, and insightful information to support effective decision-making across the organization. Many roles have some level of manual reporting required, which typically involves extracting data from IT systems and cross-referencing it with other available data.

For example, a large finance team in a consumer goods division for a multinational identified that they spent the equivalent of over eight employees’ time on generating various types of reports, accessing over 20 different data sources. The team generated over 30 reports in response to requests from different parts of the business, updating marketing on their cost of sales for specific product lines, and HR on labor-cost spend. The team also spent a considerable amount of time creating standard reporting decks to serve various decision-making forums, such as weekly operations management meetings.

While the organization had only recently initiated a wider transformation program, which included a data strategy component and a move to self-service reporting, it was expected that this would be at least 18 months away from delivering any specific changes to working practices. As an interim, the financial team conducted an internal piece of analysis and rationalized its current reporting stack by over 30%, before building the requirements to automate the various processes. As well as creating and issuing the reports by email on a more regular cycle, the digital worker was also responsible for creating draft reporting decks, including data sets for issue by email to internal teams.

A European-based telecom provider regularly changed its pricing plans based on new offers for broadband, fixed-line, mobile, and TV packages. The marketing team was required to check a number of complex product permutations and combinations, manually validating key factors such as bundled margins and error states by manually reviewing hundreds of lines on a spreadsheet and inbuilt macros (local spreadsheet programs).

An automated solution executed the existing validation checks as well as adding additional tasks, such as comparing the pricing of bundles with that of the firm’s key competitors. This was something that the team previously had the aspiration, but never the human capacity to do.

For example, a company within the agricultural sector was planning to decommission its existing inventory system in favor of a new system, which increased accessibility to its wide supplier network. However, due to recent growth, it was spending increasing amounts of manual activity ensuring alignment between its inventory and financial systems. The business was unable to provide accurate, regular updates on the value of technical agricultural equipment stock held, which consequently delayed spend on the production of new equipment and sales.

In another example, a telecommunications company had completed due diligence on a new, additional software component for its existing CRM system. It would automatically issue text alerts when an engineer had been scheduled to visit a customer. The cost of this additional software, including integration, was quoted at £420k and would have taken 10 months to implement. Fortunately, prior to implementation, the business configured a digital worker to provide the same functionality by receiving the booking event from the CRM system before sending a text message and updating the customer’s record.

The customer services team was able to reduce its budgeted spend by the amount initially allocated to the original solution, but more importantly, deliver this improved experience to its customers in a fraction of the time. The speed at which automations can be deployed can provide a significant advantage.

A large construction company, based in the US, used a number of different agencies to source resources, from engineers to construction workers. Each of the subcontractors was required to complete a timesheet containing the number of hours worked per week and an approving manager’s signature. Over a two-year period, the organization accumulated nearly 100,000 timesheets, which were archived onto a storage device within the corporate systems in a PDF format.

A manager within the organization identified a specific discrepancy between an agency’s invoice and work completed. It highlighted the fact that some of the workers’ rates had been applied incorrectly. With top-level support, the manager was provided with a digital worker that extracted key data from each of the timesheets and placed it into the enterprise data repository. With the data now accessible to analysis, the organization identified over $2m worth of overbilling, due to rates being inconsistently applied, and progressed refunds with the largest agencies.

Automation provides a wide variety of different benefits to the organization. As the enterprise’s automation capability matures and scales, the latent potential of the technology increases, supporting the application of the technology to wider business challenges. In turn, this unlocks even greater value and benefits across the organization.

Process automation can make a process more efficient by replacing activities or tasks that were previously completed by employees with digital workers, freeing up the capacity of resources. Creating this spare capacity in teams can be used in a number of ways to manage costs, for example by removing the need for additional hires and refocusing existing capacity on new tasks, or by directly reducing the size of the teams. Depending on the complexity of the process and the number of resources currently required, it may also free up time from indirect roles, such as team leaders or supervisors, further improving efficiency. Business processes contracted out with third party BPOs (Business Process Outsourcers) can be reduced, insourced, centralized, and automated to directly reduce spend.

In contrast, a digital worker follows its sets of work instructions without deviation, 100% of the time. If an error is found it can be fixed in the underlying program and redeployed so it does not occur again. Less process “stop” time means more of the process is working in an automated state, which can in turn lead to increases in productivity.

Automation may increase throughput speed, allowing the time within which the process cycle can be completed to be reduced. One of the fundamental benefits of digital is elasticity. In this case, the elasticity of capacity allows potentially any number of digital workers to be applied to a process in order to reduce cycle times. This leads to the potential of improving SLAs (service level agreements) and responsiveness to customers, providing the opportunity to significantly improve performance compared to manually transacted processes. Technical limitations, such as the responsiveness of certain IT systems, can limit the process cycle time speeds, which can be worked around using the flexibility that automation can provide. For example, a recently deployed digital worker logged into an IT timesheet system daily to upload time and task information from a separate system. On launch, it was identified that the digital worker’s full capability was limited as it was scheduled to log onto the timesheet system at 8:00 a.m. every day to execute its activities—one of the busiest and slowest times of the day for the system. The digital worker, like a human employee, would observe a system responding slowly. So the automation was rescheduled to start at 3:00 a.m., which reduced the cycle time by over 70%. Importantly, it also allowed team members in HR access to key reports by the start of their working day.

Automating the process effectively standardizes it. Where previously there was “wiggle room” to deviate, automation places the process into a straitjacket, providing direct alignment between the corporate strategy and the operational execution. While this standardizes the activities the digital worker completes, it more importantly demands further standardization by the employees and systems that work upstream and downstream of the automation. Typically, they will need to “feed” the digital worker with standard inputs and receive standardized outputs from the automation. This creation of a standardized ecosystem in turn drives both improved productivity and quality.

For example, a global logistics company implemented a digital worker to reconcile daily inventory stock levels of its German operations. Prior to the automation, employees would capture this information onto a set of spreadsheets and email them to a colleague to reconcile the spreadsheets manually before entering them into their IT system. The variability of the process created significant discrepancies during month end, impacting the accuracy of their balance sheet.

The automation calculated what should be in the working inventory from supplier invoices and validated this against the now-standardized inputs being received from employees. If the input information was not received on time from the employees, an email was automatically sent, which notified them and provided a revised window. As a result, the inventory digital worker was incrementally rolled out to all other European sites, driving adoption of the same process to drive standardization across all regions.

One of the benefits of a programmed set of work instructions to be followed by the digital worker is the full adherence to procedures. A process standardized for automation provides a very clear set of rules that need to be followed. While following these procedures, the digital worker completes a full audit trail. It logs every system it has interacted with and every data item manipulated, significantly improving the compliance and also the auditability of a process. This capability can be used to regularly monitor environments to ensure compliance, flag any risks to statutory and regulatory requirements, and minimize the risk of fraud.

For example, for a new banking business account process, a digital worker can validate a number of details following the submission of an online form. It can verify the company’s trading name using a third-party service, check the address and postcode, and carry out a credit check using a credit agency’s portal. The digital worker can own the complete end-to-end process, only breaking out of the account boarding workflow when certain business rules are breached—for example, if the customer’s details do not match the address information provided.

Importantly, to comply with Sarbanes-Oxley risk management and audit procedures, the digital worker will have created a complete audit trail of each of its steps and injected this data into the bank’s database system.

Automation provides the technological opportunity to take repetitive, mundane activities off employees’ to-do lists, directly improving their level of satisfaction and engagement. The Bain Global Automation Survey, conducted in 2019, used data from over 790 interviews to show that organizations ranked automation’s ability to free their employees to do higher-value work as the most important benefit. [39]

A benchmark survey by Forbes Insights with 302 senior executives highlighted that “92% of respondents indicated improvements in employee satisfaction as a result of their automation initiative. More than half, 52%, indicated employee satisfaction increased by 15% or more.” [40] Automation provides an unprecedented opportunity for organizations across sectors to transform how their employees work, that not only provides greater alignment to the corporate strategy but also can significantly improve employees’ level of satisfaction in their roles and provide skill-based growth by liberating them from repetitive, boring tasks to engagement through high value-added activities.