In our previous post we looked at the global state of outsourcing and BPO, the signs of its transformation, the appearance of Robotic Process Automation (RPA) as a new (?) emerging technology trend and its market size and benefits. Today we’ll dig deeper into RPA examining its effect on workforce transformation (more precisely on BPO), highlighting its limitations and through that look at its role in the Data Capture and Data Extraction and Transformation industry.
Impact of RPA on employment and workforce
Looking at the benefits gained by implementing robotic process automation, and the current market trends, can we say that RPA will replace traditional outsourcing in the IT sector? Will all those people, working in the BPO sector (or even in on-shore business process management), loose out their current job to software robots? What effect will RPA have on workforce altogether?
Some academic studies project that RPA, among other technological trends, is expected to drive a new wave of productivity and efficiency gains in the global labor market. Although not directly attributable to RPA alone, Oxford University conjectures that up to 35% of all jobs may have been automated by 2035.
At the same time, according to Harvard Business Review, most operations groups adopting RPA have promised their employees that automation would not result in layoffs. Instead, workers will be redeployed to do other, more interesting work. Thus, the adoption of RPA won’t mean a smaller headcount, but rather a way to achieve more work with the same number of people.
Another academic study strengthens this idea, showing that knowledge workers did not feel threatened by automation: they embraced it and viewed the robots as team-mates. Quite contrary other analysts proffer that RPA represents a massive threat to the (BPO) industry. The effect, if true, will be to create high value jobs for skilled process designers in onshore locations but to decrease the available opportunity to low skilled workers offshore.
Let’s just say, this discussion appears to be healthy ground for debate. However, what we can already see, is that large global IT and business process outsourcing companies – ones mostly effected by the rise of RPA – are quick to react, by adopting robotic process automation to strengthen their position rather than fight against it. Of course, some IT service lines will come under RPA’s influence and will probably be fully automated.
Repetitive, rules-based business processes, such as payroll processing, data collection, or application support and maintenance, including activities such as monitoring, providing clarifications and reporting or common service requests such as password resets and users’ requests to access systems. These are all typical BPO tasks, and given that software robots can be 50-70% cheaper than an offshore full time employees BPO’s are quick to react to this threat
- by incorporating RPA into their own processes (before their clients will do so) and thus creating even more competitive offerings,
- by combining RPA with cognitive technologies improving RPA capabilities themselves
- by looking beyond, the simple model of cost reduction and focus more on value creation capitalizing on RPA’s limitations
Until RPA can entirely automate all BPO services or until other automation platforms such as AI and cognitive become more commonplace, BPOs will be here to stay.
Limitations of RPA
We can clearly see the effects of RPA on modern workforce, but does it mean that RPA will replace human workforce in every business process related job in the near future? The short answer is no (at least for a few more years).
For sure, RPA is an evolution step in computer related process automation, but it might not be so groundbreaking as some RPA vendors like to portray it. The work of “intellectual workers” (in a very wide sense of the word) is still a long way from being fully automated. RPA is designed to deal with repetitive tasks, that don’t require “thinking” and can be defined by a set of well laid out rules covering all possible scenarios. There are still clear limitations to the types of work in which these robots can be effectively utilized. Software robots are bad in things like, judgement, perception or something that’s so simple to us, humans – natural language processing. This means that processes that require human judgment within complex scenarios—for example, complex claims processing—cannot be automated through RPA alone.
Of course, this can be improved by combining RPA with cognitive technologies, but that’s a whole new topic to discuss. Additionally, to work properly, these software robots require relatively clean data, which can be a rare thing in today’s business processes. For example, a lot of business processes still happen on paper – a data source robots are not capable of processing well, especially when we’re talking about semi-structured on unstructured formats and handwritten text.
In the end, robotic automation still has significant limitations, so it’s essential to identify which processes can be performed well by robots and which are not. Robotics should be seen as just one component of end-to-end process improvement not a universal, complete solution.
RPA in Data Capture and Data Transformation
Despite the limitations described above, many RPA vendors include data capture, data entry and data transformation as one of those IT segments, where RPA will revolutionize workforce and improve processing. (Vice-versa, some capture and data extraction vendors incorporate RPA into their offerings). Is this real or are they just riding the RPA bandwagon? Let’s take a closer look.
Although the general usage of paper is globally declining, in some industries becoming “paperless” is a much slower process that in others. These “paper-heavy” industries typically include the financial sector (banking and insurance), the government sector and healthcare. There are many reasons to this, including the involvement of handwriting (due to security reasons and data confidentiality), the slower reaction to emerging modern technologies due to their size and rigidness, or just the insistence to old habits and workflows. Thus, in many cases paper usage is not declining, but stagnating, or in some cases are even rising.
If we take a step back from RPA and look instead at automation in general, we see that it’s been in use in the data capture and paper processing industry for quite a while. The first introduction of automated recognition – the omni-font optical character recognition happened as early as the late 70-s (courtesy of Kurzweil Computer Products). By the late 90-s and early 2000-s more and more vendors appeared specializing in automated optical character recognition, irreversibly transforming traditional paper processing.
From the traditional, full manual data entry – where users were directly working with the paper source and typing data into a text file – by now, we reached an evolutional stage where scanned documents are directly imported into special capture software where they are cleaned up using image pre-processing, automatically identified and classified using various technologies that try to mimic human understanding (fingerprinting, anchor point-based classification, key-word based classification, etc.) and then automatically processed, or more precisely recognized by special automated character recognition engines. These engines can be categorized into 2 main categories: OCR (optical character recognition) engines and ICR (intelligent character recognition) engines. The first one deals with machine printed or typed text, while the latter is used to recognize handwritten text.
Naturally, with the character recognition engines it’s all about speed and accuracy and it’s no surprise that they’ve improved a lot since the early days. However, the progress is slowing down. Engines vendors like to state that their automated recognition rates are 90%-95% or even 99.9%, but this is just over the top marketing and in the real world, these numbers are much-much lower. Especially when it comes to handwritten text recognition (ICR). The truth is that data validation (supervision or quality control – you name it) is still done manually by human users and is still an essential part of the processing. This manual validation can be optimized, minimized, but it can’t be fully automated or “robotized”. At least, not yet. Also we can see, that all other above-mentioned steps of the processing are already designed to be fully automated. Drawing a parallel with RPA, we can say, that these software solutions in a way already function like document sorting and data entry robots mimicking human users. There isn’t much that RPA can “revolutionize” here. Based on this, we can say that the RPA technologies we know today will have a very limited effect on the current data capture and data extraction and transformation industry and market.
However, there are some very interesting new trends and technological approaches that will most likely shape the industry in the coming years, of which we’re going to talk about in the upcoming posts.