What is RPA? Why is it important? What are its benefits? What are different types of RPA? How does it work? How to choose RPA vendors? How is RPA priced?
Movies have been feeding us physical robots for a long time. However the first AI bots that we will work with are software bots. This shouldn’t be surprising as hardware is hard while software is eating the world.
Interest in RPA has grown ~10x in the past 1.5 years. This is no surprise since Fortune 500 CEOs can not stop talking about it. As John Cryan, CEO of Deutsche Bank said in September 2017:
In our banks we have people behaving like robots doing mechanical things, tomorrow we’re going to have robots behaving like people
With so much interest in the topic, consulting companies like BCG, McKinsey and Accenture have put together their RPA offerings. Here’s an Accenture video that outlines on a high level, how RPA works. It is a bit full of cliches but it is currently one of the most popular videos on the topic.
What is RPA?
According to Wikipedia:
Robotic process automation (RPA) is an emerging form of clerical process automation technology based on the notion of software robots or artificial intelligence (AI) workers.
While Wikipedia’s description is accurate, it could be more specific. RPA is a generic tool to create specialized agents which can automate clerical tasks. RPA is currently one of the most popular Artificial Intelligence application areas as it allows companies with legacy systems to automate their workflows. Since most large non-tech companies rely on legacy systems, there’s broad interest in RPA systems.
Why is RPA relevant now?
Because it is necessary for automating modern office tasks. We explain why in detail:
1- Employees today use too many tools to automate with simple macros
There’s an app for everything today. Hosted in the cloud, integrated via APIs, CRM, ERP, productivity and other apps run today’s enterprises. So why do we need to build our specialized robots? Well, the problem is business processes. Business processes need input from different tools and we have been using an increasing number of tools every year:
- In 1990, Office 1.0 had 3 products: Word, Excel, Powerpoint.
- Office 2016 has 9 products: Word, Excel, PowerPoint, OneNote, Outlook, Publisher, Access, Skype for Business, Visio Viewer.
With more tools, comes need for integration. Though every major tool has an API, developer company may not provide open access to the API. Without API access users are forced to export complex CSVs and resort to other complex but boring data janitor work. Who can blame the vendors? Every vendor wants to have a sticky product. Sadly. making integration and data migration difficult increases switching costs and creates frustrated but loyal customers.
Sometimes vendors aren’t the ones to blame. Most large companies still use some systems built with 2000s technology. And legacy systems are not supposed to be providing integration to modern tools. RPA provides a solution to these problems. Robots step up to bridge the gaps between systems. Additionally, some business logic involved in business processes are quite simple and bots also automate such reasoning.
2- Outsourcing no longer creates benefits as most benefits of outsourcing have been reaped
In the past, the way most companies handled the overhead required to carry out rote tasks was to simply outsource them. They received the same result without remotely the same costs.
While that was once a massive advantage, it’s fading. According to Deloitte’s Global Outsourcing Survey, in 2016, 75% of organizations profiled reported that they had already realized cost-saving targets by leveraging labor arbitrage. If that number is even remotely indicative of the larger population, outsourcing simply isn’t the same game-changer it once was.
Companies are scrambling to hire RPA consulting firms while their competitors are still relying on old-fashioned outsourcing (and the overhead that goes with it).
3- This puts companies that do not use RPA at a disadvantage
RPA isn’t going to be a luxury for much longer.
While some companies may still be competitive without it, that time is passing. RPA deployments are growing at an explosive rate. That trend is not slowing down and it’s certainly not going to reverse.
Is RPA ready for production?
Yes, most large non-tech organizations that rely on numerous systems including legacy applications already piloted RPA deployments with satisfactory results. As a result, global CEOs and their consultants are very excited about RPA:
Just take a look at these quotes from the most reputable names&brands in business:
In our bank we have people doing work like robots. Tomorrow we will have robots behaving like people. It doesn’t matter if we as a bank will participate in these changes or not, it is going to happen.
RPA is a promising new development in business automation that offers a potential ROI of 30–200 percent—in the first year
The relationship between technology and people has to change in the future for the better, and I think RPA is one of the great tools to enable that change
Leslie Willcocks, London School of Economics professor
How it works
As industrial robots transformed the factory floor, RPA bots transform back offices. RPA Bots replicate employee actions like opening files, inputting data, copy pasting fields in an automated way. To set up an RPA bot, it isn’t required to know programming. There are 4 ways to setup RPA bots
As expected, the most powerful interfaces for programming bots is a programming language. However, using a programming language requires skill and patience so this method is relevant for technically inclined personnel. Programming instructions essentially tell the bot which programs to use and how to interact with those programs.
2- Graphical User Interfaces
Many vendors offering solutions to program RPA bots with drag&drop interfaces. Anyone in the company should be capable of setting up simple bots.
Just like macros in excel, bots can complete recorded actions. Recorded actions can involve numerous enterprise software such as taking data from Salesforce, merging it with a report from mailchimp in excel to identify which customers to target during the company’s routine customer activation SMS campaign.
Recorder function is an important advantage in an rpa tool because it enables rapid bot programming. However, recorders have some limitations as well:
- Recording a complex set of functionality can be difficult and error prone
- Maintaining recorded bots is difficult as their code is machine produced and may not be easy to read. Re-recording actions after each small change in the process can also be time consuming
4- Self-learning bots
These bots watch recorded employee activity to learn automatable tasks. They are the easiest to deploy bots. However, their learning is not always perfect since they rely on recognizing images in scraped screenshots. Especially during initial deployment, they could be making mistakes and their activity needs to be audited. Most of the time mistakes are avoided as bots understand when they see cases they don’t know how to complete. In such cases, they contact employees for guidance.
Once bots are setup, an orchestration module helps start/stop bots and analyze their activity.
Deloitte UK has a nice video on how RPA works on a simple example that shows the bot extracting and processing data from an email:
Possible activities of RPA bots
RPA bots can use the operating system applications like a human user. Bots are capable of
- Launching and using various applications including
- Opening emails and attachments
- Logging into applications
- Moving files and folders
- Integrating with enterprise tools by
- Connecting to system APIs
- Reading and writing to databases
- Augmenting your data by
- Scraping data from the web including social media
- Data processing
- Following logical rules such as “if/then” rules
- Making calculations
- Extracting data from documents
- Inputting data to forms
- Extracting and reformatting data into reports or dashboards
- Merging data from multiple sources
- Copying and pasting data
Bots can do these functions on virtualization solutions like Citrix or on Windows environment. Most vendors do not support other OS environments like Mac OS or Linux. This is because most office work is conducted on PCs.
Types of RPA
There are 2 types of RPA automation that serves different needs:
These bots reside on the user’s machine and are invoked by the user. They are appropriate for tasks that are triggered at programmatically hard-to-detect points. For example, a customer service rep will understand the customer’s inquiry and need to complete a transaction in the system. Let’s assume that due to system limitations, customer service rep would normally need to work with 3 screens and complete 5 manual steps to complete this transaction. Instead of doing those, rep launches the attended automation code. RPA bot works like the rep, performs the necessary operations and asks for guidance from the rep if needed. RPA bot can actually work a lot better than the rep, perform regulatory and compliance checks and would never do manual mistakes due to fatigue or boredom.
Launcher for RPA can be setup in 3 main ways to facilitate employee’s access to the tool. Launcher can be
- on an RPA client tool where the customer service rep select the bot to be launched
- embedded on the personnel screen when certain conditions are met (e.g. when the rep is talking to a customer)
- auto-run when certain conditions are met. For example if some KYC check needs to be performed on newly acquired customer phone numbers, bot can be launched as soon as the phone number field is filled. In this manner, RPA bots can be launched automatically with no intervention from the employee.
Attended automation is a good way to augment your employees that face customers but still need to complete manual work.
Unattended bots are like batch processes on the cloud. They complete a data processing task in the background. They are ideal for reducing work of back-office employees.
There are a few options for launching unattended automation:
- Data input in a specified location: Most unattended bots are triggered when data is input in the system. Whether it is new transactions or employees, additional data processing is generally required to serve regulatory or marketing-related needs.
- Bot initiated: A bot can also launch another bot. This can be useful when a bot operation has various different outcomes. For example, a KYC inquiry may either require manual investigation or automated processing to complete the customer’s registration. Based on the outcome, bot can notify the investigation team or launch another bot to complete registration.
- Orchestrator initiated: RPA administrators can use orchestrator software to stop or launch bots.
- Specified intervals: Bots can be launched at specific times to batch process data.
No wonder interest in RPA is growing so fast. Manual processes are inefficient, prone to errors and lead to employee dissatisfaction. With RPA companies can
- increase speed of/reduce errors in customer-facing processes to increase customer satisfaction
- allows employees to focus on higher value-added activities improving both business results and employee satisfaction
- reduce manual data edits, increasing quality of data and reducing compliance risks
For a more comprehensive list, I recommend you to take a look at our comprehensive list of RPA benefits.
Essentially RPA allows a higher degree of automation through software. Of course RPA are not the only means to achieve automating for processes that cut across numerous systems. Before RPA, companies relied on 3 approaches. These approaches are shown below by Deloitte:
In short, none of these approaches offer the flexibility, speed and cost advantages of RPA. However, as CIOs know management is the art of trade-offs. And sometimes IT transformation solutions though they are slower, generate better returns in the long run. If you are not sure that RPA is the right process automation solution for your business.
Industries that are being transformed by RPA
If you ask the vendors, they will tell you that any industry is ripe for RPA automation, which is technically correct. However, RPA can have greater impact in some industries than others. RPA is a solution you should put at the top of your company’s agenda if you business fits any of these descriptions:
- Uses legacy systems
- A large portion of the workforce works in the backoffice in non-tech functions
Some industries that have companies that fit both of these points are listed below. Most of these are old companies that rely on legacy systems. RPA can achieve significant savings and customer satisfaction increase in branches, call centers and the backoffice.
Financial services including banking and insurance
According to McKinsey Global Institute’s 2017 report on automation 43% of these jobs are automatable. This is because data entry and processing is an important part of these businesses. Furthermore, these businesses are subject to constant changes in regulation such as KYC requirements. Bots can be taught regulatory changes quickly and in a centralized way. This helps companies avoid embarrassing compliance issues.
According to Accenture’s report on insurance process automation, some good activities to automate are:
- Sales processes
- Updating sales scorecards to agents
- Conduct required regulatory and legal checks
- Conduct credit checks
- Account maintenance services not currently supported by straight-through processing
- Data entry for clearance and registration processes
- Update systems with client information
- Generate a renewal premium
- Policy servicing
- Update to customer information including bank account details
- Reject or cancel policies if payments are not received
- Identify and reconcile policy premium discrepancies
- Claims processing
- Process claims payments for pre-approved amounts
- Assign to claims handlers
- Input First Notice of Loss (FNOL) submissions
- Notify loss adjusters
- Automate daily bank reconciliations
- Process low-risk payments
Utilities like telecom and energy
According to previously mentioned McKinsey report, 44% of activities can be automatable. Since these are the oldest subscription businesses, they have frequent payment and customer service requirements which can be automated.
RPA use cases/application areas
Numerous business processes have been transformed thanks to RPA. Some examples that exists in almost all industries are are:
- Application processing
- Data migration and entry
- Periodic report preparation and dissemination
How to implement RPA at your company
In short, you need to select the best candidate processes for RPA, get management and team buy-in and implementation, run a pilot and go live. While the overall process is simple, devil is in the details.
Understand the overall timeline to manage expectations
Based on the customer survey of RPA provider,UiPath, an RPA deployment typically takes 1-2 months including time required to configure, test, and launch automations into production. However complexity of the process, team size and level of automation will all impact project duration. Our recommendation would be to start with modest aims.
Plan implementation – Our roadmap based on best practice implementations
Any complex process can be broken down to simple steps and an RPA implementation is no different. Though you may need to coordinate both company and vendor resources to achieve RPA automation, knowing the necessary steps makes implementation easy:
- Select impactful yet easy to automate processes
- To maximize impact of RPA, identify impactful processes. These processes tend to be
- Impacting both cost and revenues
- High volume
- With low fault tolerance
- Error prone
- Requiring irregular labor
- Distributed processes that require coordinated efforts of multiple departments
- Select processes that can be easily automated with RPA. Such processes tend to be
- Rules based
- Not on the roadmap for new systems
- Convince the organization: Even in automation, it’s really about the people.
- Get management buy-in
- Establish governance structure
- Get team buy-in
- Implement the solution
- Choose your RPA tool
- Decide whether to outsource RPA development to an implementation partner
- Choose your partners. This includes the RPA technology provider but can also include a consulting or BPO company if your internal resources can not dedicate the time necessary to automate the process
- Run a pilot:
- Configure the RPA bot
- Test RPA bot
- Run a live pilot
- Evaluate pilot results
- Go live
- Design the new, bot driven process
- Clarify roles and responsibilities
- Go live
- Analyze results
Choose the right type of RPA tool
There are 3 major types of RPA tools:
- Programmable RPA bots: Programmers need to understand and code a set of rules governing how the RPA bot will function
- Self-learning solutions: Using historical (when available) and current data, these tools monitor hours of employee activity to understand the tasks completed and to start completing them after they have reached enough confidence to complete the process
- Cognitive/intelligent automation: Cognitive automation solutions (also called smart or intelligent automation) self learn and deal with both structured and unstructured data.
From the descriptions, it is obvious that cognitive/intelligent automation solutions are the most attractive ones. However, we have seen cases where programmable bots were recommended by RPA resellers rather than cognitive automation solutions.
One complication in the RPA landscape is that most companies purchase RPA solutions from re-sellers rather than the companies building the technology. Resellers allowed vendors like Blue Prism and WorkFusion to have quick access to a global customer base. Professional services companies like Deloitte, technology consultants like Accenture, Business Process Outsourcing (BPO) providers like Genpact have existing relationships with a large number of customers in various industries. They used these relationships to sell RPA solutions. And depending on the revenue model, resellers can be incentivized to increase their billable hours and recommend programmable solutions rather than cognitive automation.