The automation market, particularly the subsection of robotic process automation (RPA) designed to reduce the burden of repetitive and simple tasks, has recently been picking up considerable speed. Companies, such as UiPath and Automation Anywhere are drawing mass amounts of venture capital funding and have grabbed the attention of tech giants looking to expand their stacks — prime examples being Microsoft’s recent acquisition of Softomotive and IBM’s purchase of WDG this month.

Major changes in business operations since March — brought about by the shift to a remote workforce and the need for social distancing — mean that automation of business processes will likely become even more critical going forward.

However, despite the influx of capital into RPA ($6.7 billion according to Pitchbook) and Gartner reporting that it is the fastest growing market in enterprise software, automation should not be considered a silver bullet solution, and real life implementation has struggled to meet expectations to date.

The COVID-driven move to automation

The current global pandemic has shifted all industries to remote-first, with business continuity becoming the No. 1 priority.

While automation initiatives were initially put on hold at the start of COVID, this new world has shone a spotlight on major gaps in existing processes, creating a greater need to optimize resources, accelerate company performance, and improve business resilience. As Microsoft CEO Satya Nadella said in April, because of COVID-19 “we have seen two years’ worth of digital transformation in two months.”

The pandemic is now considered an inflection point for automation, with some predicting that RPA spend will reach $25 billion by 2025 (compared to $3.6 billion today). In the first quarter of 2020, UIPath added 836 customers, as it helped a number of enterprises ensure business continuity, including patient data collection, health alerts, government approval for stimulus assistance, contact tracing, and more.

While automation can be a game-changer during this critical time of accelerated digital transformation, it is not the answer for everything. We’ve been speaking to a vast array of Fortune 500 executives lately to get the top takeaways for other companies considering a move to process automation.

The process of automating

Successful implementations of RPA start with concrete problems and use cases where automation can be implemented (Stage 1). One of the first challenges is understanding what kind of automation will solve your specific use case or business process. The key lessons here are:

  1. Focus on the long-term objectives of each business process and the entire chain of command rather than on individual tasks.
  2. Start with business processes that are less complex yet highly strategic to build early success internally and externally (e.g., invoice processing, credit checks or inventory management). Don’t be discouraged by underwhelming results, as it is an iterative process.
  3. Align the strategy between management and employees, as many employees do fear replacement. With both parties aligned, identifying bottlenecks and overall productivity improves.
  4. Security. Security. Security. Establish governance and work closely with IT during the deployment phase to reduce the long-term risks of an increased surface area.

Most corporations we spoke to have selected either UiPath or Automation Anywhere for RPA. During deployment (Stage 2), we find that most enterprises start with back-office functions, mainly in finance and supply chain. Business processes within these two departments are complicated, requiring input from multiple departments in both structured data (from Salesforce, ServiceNow, SAP, Oracle, Excel, etc.) and unstructured data formats (email, pdfs, hand-written notes, etc.). Therefore, many choose to start by automating individual repetitive tasks and simple workflows that mainly work with structured data formats.

As users gain confidence in RPA (Stage 3), several corporations expand their usage by deploying more RPA in increasingly complex workflows and adopt additional tools. These may include business process management (BPM) platforms with a built-in integration platform as a service (iPaaS), optical character recognition (OCR) tools, or conversational chatbots. It’s important to recognize the limitations of RPA and the growing value in outside solutions. For example, one of the most significant constraints to RPA is that the technology does not work well with unstructured data today.

Through this journey, enterprises begin to take a different perspective on their automation strategy and solidify best practices for future deployments (Stage 4). Some revisit their existing internal processes, generally with the help of a process mining or discovery tool. With better visibility in their internal workflows, executives can evaluate the necessity of the RPA already deployed and the potential need to re-engineer or redesign the process. A few trends we have seen with corporates at this stage are shifts in focus towards:

  1. Purchasing end-to-end (E2E) business process verification and orchestration platforms to highlight potential bottlenecks and to empower end-users to design processes with modules powered by machine learning or natural language processing (NLP). Of note, several organizations saw significant improvements in the ML/AI capabilities on unstructured data formats.
  2. Leveraging process mining tools as the management layer for both work and automation. Executives are starting to understand that automation is not always necessary in certain processes and that bottlenecks can be resolved through other means such as a redesign in workflow, retraining of employees, or modifications to the system.
  3. Increasing architecture flexibility to enable API driven processes and improve overall agility. Of note, most organizations do recognize that this will likely take much longer.

There’s no cut and dry definition of RPA success; it will vary depending on a company’s unique workflows, and there are some use cases that are better suited for RPA than others. At the highest-level, RPA is about re-examining how a business runs and determining if things can be re-engineered to achieve greater efficiencies. Put simply — it’s all about optimization.

Top tips for implementation

If you’re new to RPA, here are the top tips for success we gleaned from the Fortune 500 execs we spoke to:

1. Align your workforce: The first step is establishing alignment across departments. There is real fear from employees when they hear the term “automation” — a recent survey shows that nearly 25% of employees fear losing their jobs — so ensuring employees understand the objectives (e.g., empowering them to focus more on mission-critical processes) is vital for long-term success.

2. Implement governance and prioritize security: Create a Center of Excellence (COE) that operates as the “HR” of automation in your organization to govern what works, where automation is needed, and best practices for security and IT. The COE can also provide change-management for exception handling and process overview, guidance, and course correction. The COE manages the actions of the automation tools, similar to how HR manages human labor. This will ensure that “shadow IT” does not occur and actions taken are in accordance with compliance and security.

3. Focus on visibility, documentation, and lastly automation: As companies stay remote longer, they’ll need to better understand their existing internal processes, with documentation being key for data-informed decisions on the best path forward. Catalog internal processes with process mining tools before selecting areas to automate. Historically, process mining was log-based (Celonis) to visually digitize workflows, but now computer vision-based tools (FortressIQ and Skan.ai) provide increased accuracy in revealing mismanagement, inefficiencies, compliance issues, and the ineffectiveness of existing tools. Automation is not always the answer, with alternative paths including redesigning workflows or retraining employees.

4. Target long-term objectives, not individual tasks: After documentation, change-management must think critically about whether automation is truly needed and, if so, what improvements it can offer. Process intelligence is important so companies can think through from a first principles perspective to re-engineer their process, re-train employees, or implement automation. A business must first understand its existing processes before considering implementing automation technologies.

5. Juggle best-of-breed and horizontal tools: Most executives recommend using both horizontal and best-of-breed solutions. End-to-end intelligent business process automation platforms (e.g., TonkeanAutomation HeroCamundaWorkfusion) allow companies to consolidate tools and reach the desired business outcome. These are especially helpful as most business processes include a combination of human intervention and data manipulation, with complex workflows consisting of numerous data formats and applications.

However, automation can be brittle as processes change over time. Industry-focused companies such as supply chain (Slync.io), finance (Ocrolus), and healthcare (Notable HealthOlive AI), or departmentally focused solutions such as accounts payable (Stampli) and user interface (MesmerHQ), can contextualize data and provide higher accuracy given their specialization. The customized workflows tend to be easier to use than horizontal solutions where end-users likely have to start from scratch.

Approach RPA iteratively

To achieve the best results with RPA, take an iterative approach. Don’t expect perfect automation after a 12-month development cycle.

Successful deployment or implementation occurs when the automation is repeatable and scalable as work becomes more complex. As the automation project shifts from lower-level, mundane tasks to significant functions such as orchestration and resource allocation, this is a strong sign that automation is augmenting the productivity of the team. When successful, business objectives leveraging automation can affect both top-line and bottom-line results.

The COVID-19 health crisis is rightfully on everybody’s mind. However, once it passes — which it eventually will — one of its less obvious lasting legacies will be how it was the catalyst to mainstream adoption of automation.

Simon Wu is Investment Director and Jason Chen is a Venture Investor — both at Cathay Innovation.


The audio problem: Learn how new cloud-based API solutions are solving imperfect, frustrating audio in video conferences. Access here