RPA 101 – we are assuming nothing and are beginning the journey at the very start.
Robotic Process Automation (RPA) can play a significant part in a business’ digital strategy. It can unlock value that sets your company apart from your competitors. It is an exciting technology that is drawing increasing amounts of IT, analyst, consulting and business attention. In this article, we will go back to the very basics to help everyone who is new to RPA.
Badging everything as Robotic Process Automation is definitionally incorrect (e.g. Desktop Automation; Attended, Unattended or Hybrid Automation). However, for simplicity sake, we will use RPA as an umbrella term to describe process automation software technology that attempts to mimic what humans do at the computer desktop application layer (i.e. the computer operating system and programs that you and I see).
What is Robotic Process Automation (RPA)?
RPA is an emerging form of business process automation technology. It is used to mimic repetitive human tasks. It is ideal for those processes that do not require cognitive (human) decision-making or repeated human intervention. Processes with yes/no decision points are all perfectly suited to RPA. RPA software interacts with user interfaces (computer programs on PCs screens) much like one of your employees might. RPA software is used to build computer code. A bot, also known as a robot, or digital worker, is a software application that runs scripts that automate processes or tasks.
Deloitte have created an excellent YouTube video that shows RPA taking data from email attachments, ingesting the data into an accounting application and emailing confirmation to clients. (This is not a recommendation for Deloitte; the video is simply excellent).
When does RPA work best?
On paper RPA works well when; processes are pre-defined; structured; repetitive; easy to understand; digital; high volume; where data quality is excellent and there is a logical set of predefined steps to follow. RPA does not work well when innate human judgement is required. Such processes are best left for people to complete. It is arguable that automation programs struggle with technologies such as iframes and Citrix but strides are being made by many vendors to overcome this challenge.
“Before diving into your RPA journey, you need to be certain that your processes and people are ready. If you select the wrong process, then you will not succeed“.
Matthew Coffey, RPA Delivery Lead at Pearson
Practically, RPA works well when an organisation is clear as to what it wants to achieve from its digital transformation program and understands where RPA can fit in (and where it can’t). RPA needs a strong executive leader to gain executive support (and funding) for the program and it requires an organisation culture that is open and supportive. RPA programs will not scale, and most likely fail, when processes and data are constrained within physical and cultural silos.
What is the difference between attended, unattended and hybrid automation?
A robot is a robot, right? Er, not quite. There are different types of automation solutions. RPA solutions tend to come in 3 different ice cream (or is that ‘I scream’) flavours i.e. attended, unattended and hybrid (automation).
Attended automations run on a worker’s desktop and are invoked by the user via context-based screen triggers, an embedded screen button or hotkey combination, etc. An attended RPA application ‘works’ alongside agents with humans and robots working together to run a process greatly reduce the time required to perform activities. This automation type is most commonly used in front-office scenarios e.g. a call centre, a claims handling function.
This type of automation is often deployed when bots need human help to make decisions or when work is routine in nature but unpredictable in arrival patterns. For example, in an insurance firm, thousands of claims must be processed quickly and accurately. As the processing of claims tends to follow a pattern, claims tend to a perfect area of opportunity for RPA. First response telephone-based claims calls are not entirely predictable and therefore batch processing is usually not possible. When a call is received a claims handler can interact with the claimant and robot to quickly onboard a claim. Sometimes additional input or approval is needed by a claims handler in order to complete a claim process e.g. to handle a disputed claim, or potential claims fraud has been flagged and needs investigating further. Attended bots can be programmed to identify those instances and defer them to a human.
Unattended bots are deployed remotely on a server away from workers desktops and ‘supposedly’ without the need for human intervention. Unattended automation is used to automate high-volume transaction-based back-office processes attempting to remove people from the process entirely. We say ‘attempts’ not ‘will’ as dirty data can and often does cause robots to stop processing (or throw up an exception which needs human intervention to solve, much like a worker may have to raise their hand to get a supervisor or colleagues guidance). For example, a robot can be used to extract invoices attached to emails which have been sent to a company. This simple task can save a company a great deal of manual labour. However, if the robotic process has been scripted to recognise .docx extensions and the sender emails a .pdf document instead then human intervention may be required to ‘tell’ the robot what to do (in the real world this would most likely have been coded for).
Attended and unattended RPA can also be used together in a hybrid model. A task can start with an agent and be enabled by an attended automation, which can kick-off unattended RPA to complete a process.
Why has RPA has attracted so much headline-grabbing attention?
There has been massive growth in the RPA sector with many diverse tools and product features popping up faster than toast (see AI Multiple’s blog article listing 52 RPA companies). The top three companies in the sector have been estimated to have a collective market valuation of $11bn or more (Automation Anywhere, Blue Prism and UiPath). This enormous valuation is largely derived from RPA’s ‘promise’ to help enable firms to realise value from vast amounts of inefficient, manual processes with cheap ‘bot’ workers rapaciously executing costly, processes 24/7 without the need for a person (or rather the cost of a person).
“There are many benefits that can come from introducing RPA, from financial gains to better controls and improved staff moral but you need to be clear from the outset what goals you are trying to achieve and communicate this throughout the organisation”.
Andrew Hartley – RPA Consultant
RPA is exceedingly good at breathing new life into legacy processes and ageing technologies. There still exists a great number of manual processes and ageing tech within many businesses. Consequently, RPA can be applied to a wide variety of rules-based processes across a wide range of industries and sectors including; banking, education, healthcare, logistics, insurance, food and beverage, entertainment, consumer goods, retail, telephony, oil and gas, car manufacturing and many more.
Why bother with RPA?
For years businesses have attempted to reduce their costs by using business process outsourcing firms in India, the Philippines and beyond to garner the benefits of labour (wage) arbitrage. However, as their competitors similarly outsourced to compete for the need for businesses to further reduce cost, increase efficiency, and provide high-quality digital services has driven firms to take advantage of automation technologies like RPA.
“RPA is not about talent replacement, it’s about talent augmentation. – not Just about cost optimization, it’s also about organizational transformation. – not about taking the robot out of the human, but bringing the human back inside the human”.
Shail Khiyara -RPA & Intelligent Automation Executive
Bots ‘can’ work at a considerable pace (i.e. 2 to 5 or more times faster than a person). They don’t always do so due. Technical challenges associated with your network, software, bandwidth, etc. can all impact how fast bots work so do remember to test and check how fast a bot will in fact work. Bots are cheaper than a person with a figure of 1/3rd the cost of an offshore resource often quoted (we will revisit this concept of cheaper later as RPA is not cheap to implement at scale and can, in fact, cost more than a human). Bots don’t take breaks (unless we count downtime for RPA vendor updates). Bots don’t need holidays (they sound quite boring, don’t they?). Bots ‘can’ work around the clock 7 days a week, 365 days a year (remember you will have your own IT system downtime). But contrary to vendor sales pitches, bots are not cheap and when used incorrectly bots can be as troublesome as your most disengaged employee!
How come there are so many people rushing into RPA?
With billions of pounds of over-eager investor cash on the table RPA vendors have rushed to scale to meet demand, hiring sales staff that promised endless riches without understanding the complexity of their product, the customer organisations, their poorly constructed processes or their siloed cultures. RPA vendors rushed to add ecosystem partners at a rate of noughts; their products chock full of oven-ready features and promises.
Almost simultaneously analyst companies overhyped the benefits of RPA in the rush to get in on the gig (anyone know an analyst who has written RPA code?). Consulting firms began to use RPA as a catalyst to land and expand large digital transformation programs. Executives, excited by the prospect of digital transformation at scale at next to no cost and without the need for ‘pesky’ IT teams, bought into RPA in droves.
When vendors, analysts, consultants, and companies struggled with RPA the narrative changed. Suddenly broken RPA programs could be solved by adding AI (artificial intelligence) to their programs. Voila ‘cognitive computing’ was born (i.e. RPA plus AI plus Analytics). ‘Should know better’ executives, who struggled with RPA, now had the promise of digital transformation tantalising dangled at their fingertips again without the need for those awful folks in the IT department who keep saying no, and more riches were promised than before. All horse shit of the highest order but who says we never learn.
What is Cognitive Automation?
There are many situations that require human judgement, particularly when large volumes of data are involved. This is where cognitive automation comes to the rescue. RPA relies on basic technologies, such as screen scraping, macro scripts and workflow automation. Cognitive automation, on the other hand, uses more advanced digital technologies, such as natural language processing (NLP), text analytics, advanced data analytics, data mining, semantic technology and machine learning, to make it easier for people to make better business decisions.
What flavour of automation would be the best fit for my company?
If your businesses processes have structured, voluminous data and are strictly rules-based, then RPA ‘could be’ the right solution for you. However, if your organisation deals with complex, unstructured data that requires human intervention, then cognitive automation could be better suited to your organization. If your business processes are a combination of simple tasks and tasks that require human intervention or judgement (most businesses), then you can opt for a combination of RPA and cognitive automation technologies. The right answer is not necessarily choosing one or the other. As RPA and cognitive automation define the two ends of the same continuum, organizations typically start at the more basic end which is RPA (to manage volume) and work their way up to cognitive automation (to handle volume and complexity).
Vendors, are you listening to this next sentence? In the real-world business customers just want automation. End of. Firms don’t want to have to pick or decide what flavour of automation they need to get the job done. Vendors need to make the gap between unattended, attended and hybrid seamless. Businesses just want RPA to work.
It sounds easy to pick one RPA tool, right?
Again, no. Let’s assume you have a 100 person claims contact centre, you might believe that you need an attended or hybrid RPA tool. Well yes, BUT. True digital transformation should involve your reimaging your processes for the digital age. Rather than using RPA to make your existing crappy processes faster and cheaper, firms should consider redesigning processes to drive specific, digital age business outcomes.
“RPA is not about talent replacement, it’s about talent augmentation. RPA is not Just about cost optimization, it’s also about organizational transformation. RPA is not just about taking the robot out of the human but bringing the best of both to deliver exceptional value to your customers. You need to understand what tool is best for you to deliver that value and that is a very personal decision for all orgnisations.”
Imagine a scenario where a customer has had a car accident. What if the claimant had a claims app that verbally guided the driver as to how to proceed when they have been in an accident? The driver might be asked to visually (camera pictures) and verbally log a record of the incident (NLP could translate the verbal message into text which could then use AI to analyse the now text for call handling direction, fraud, etc.); pictures of the scene could be geolocated and drawn onto a digital map in office for a claims handler to view; pictures of the damage to the car could be evaluated for cost of fixing with an appointment booked the claim handlers partner repair network and sent in an email; a taxi could be ordered to take the person from the scene and a pickup truck sent to remove the vehicle, and if the call warranted a claims agent to call then a call could be placed to the drivers phone to check in with them at the point where the claims agent can add value. All this work might involve a digital (phone app) to a digital transaction (unattended RPA bot in data centre monitoring a claim inbox) with no or limited human involvement.
Subsequent to the claim settlement a Net Promotor Score (NPS) survey might then be sent to determine how the firm did during this key moment of truth. RPA could be used to send and receive the results with AI used to analyse and determine the next best action in each client instance.
“Firms need to think beyond what they are currently doing if they want to out compete your competition. Don’t just automate how you currently work. Digital has opened a host of opportunities that allow you to truly reinvent and transform how you deliver value to your customers.”
So, what next?
Companies may be tempted to simply choose a hybrid solution to be safe, but each RPA software comes with its pros and cons so do think above and beyond.
There are a lot more things you need think about beyond attended or unattended automation before you buy a tool (e.g. vendor ecosystem; availability of skilled resource; AI and analytics capabilities; unstructured data handling abilities; cloud readiness; security features, etc.).
In our next article, we will help guide you toward which solution best fits your organisation. In the meantime, email Kieran and myself your comments questions and queries. Like the marines, we wish to leave no man or woman behind before we get to the next article.
Note: for those who want a more defined definition of RPA, the IEEE Standards Association published a Standard to guide the Assessment, Evaluation, Comparison and Selection of Robotic and Intelligent Process Automation Products and Features (IEEE 2755.1-2019 Guide for Taxonomy for Intelligent Process Automation Product Features and Functionality). This detailed guide assesses the features and functionality of intelligent automation products, and defines more than 140 product features and functions, detailing their importance, and providing guidance on the assessment process.