Building a Solid Foundation for Your RPA Implementation
Ernst & Young (EY) is a strong believer in the power of RPA. The firm not only has a thriving practice in the industry but also uses the technology to improve its own operations, on a global scale.
But this does not mean that EY is not without its criticisms of RPA. The firm realizes that – as with any type of major enterprise software system – there are many potential land mines. Of course, EY’s consultants are good at navigating clients through them. Interestingly enough, one of the firm’s fastest-growing businesses is providing remediation services for failed RPA implementations!
In a report from the firm, there was an extensive analysis of a broad-range of RPA implementations (spanning 20 countries). The high-level conclusion? Well, it was not encouraging. EY found that anywhere from 30% to 50% of initial RPA projects failed.1
Now this is not an indictment of the technology. Note that EY believes that RPA vendors do generally have solid software. But as with any technology that has a potential major impact on an organization, there needs to be considerable planning. But unfortunately, this is something that can get scant attention.
In this chapter, we will highlight the initial steps of RPA, so as to put together a plan that will provide for a durable foundation.
The Preliminaries
It’s true that RPA is relatively easy to use because of the interactive workflows and drag-and-drop capabilities. But this can lead to a nagging problem – that is, it is tempting to just rush and get started with an RPA implementation, such as by creating many bots to automate processes.
On the surface, this approach has some intuitive advantages. Hey, after all, shouldn’t an organization be agile? Isn’t it a good idea to experiment and try innovative ideas?
These are all important. But if you are too eager to implement RPA, the outcome can easily be chaos and failure.
Rather, a preferred approach is to draft a written plan of action, which provides the key priorities, objectives, and roles for the RPA implementation. There should also be a look at security, the impact on IT, governance, and compliance.
As should be no surprise, a good exercise to start with is brainstorming. At this stage, be freewheeling and do not criticize ideas. The goal is to unleash creativity and ingenuity. As much as possible, try to get feedback from different parts of the organization, such as marketing, finance, HR, legal, IT, and sales.
What is the degree of automation in your department?
What works already? Why?
What is falling short?
What processes can be improved? And can automation help?
What are the processes that are repetitive and routine? How many people are involved? The time spent on the processes?
What technologies have not worked in the past? Why?
What is the general sentiment about automation? Is there resistance to this? Might employees fear for their jobs?
You will essentially be in a discovery mode. So make sure you have an app that can track the ideas and allow for commenting. Something as simple as putting together a Google doc can help with the process.
For some companies, they will set up a workshop, say, for a few days. It will be an intensive focus on ideation. But it is also an opportunity to provide some training, which would include tutorials on the fundamentals of RPA, the benefits of the technology, and how it will change daily work.
Introduction to software bots and automation: This is a general primer on RPA and how it is used. There is also coverage of different use cases that are relevant to the client’s industry. Finally, CloudStorm goes over what tasks are the best for automation as well as some of the gotchas to avoid.
Process Automation Identification and Elaboration: This involves group activities to come up with ideas to identify areas to automate. By doing this, you can have a white board or wall where you pin them. After some discussion, the team can then score the ideas (in terms of feasibility, cost, and potential return), such as by voting.
Discussion and Validation: This is another interactive discussion. You can divide the group into different teams and then present and validate the ideas.
The next step? CloudStorm will draft a report with the results of the workshop and provide next steps: Suggest areas to automate, estimate the expected impact, and craft the KPIs.
Finally, some companies will use software to help with the mapping and optimization of processes. This is something known as process mining. We’ll cover this in detail in Chapter 12.
Use a Consulting Firm?
It’s common for a company to use a consulting firm when planning for an RPA implementation. Granted, this may wind up costing more and yes, it can be tough to find a qualified consulting firm.
Or in some cases, the consulting firm may not have the right team to help you out, such as by having direct experience with your industry or company size.
Despite all this, there are certainly some clear advantages of hiring a consulting firm. In fact, the costs should be well worth it if you have the right partner. It’s also important to keep in mind that a consulting engagement does not have to be long term. For example, in the case of Symantec’s RPA implementation, the company retained a consulting firm for the first couple months so as to help with the initial planning and bot development. After this, the company’s own internal team took over the project.
The company should have experience in identifying the right areas to automate. For example, as the firm will have seen the full life cycle of an implementation, it will understand what to anticipate and the danger points. This can be a major factor in a successful RPA implementation.
You and your own team are likely too close to the company’s processes. There may actually be some glaring opportunities that could be missed! Or, in some cases, there could be areas for automation that may be far from obvious. Yet a good consultant will have a deep understanding of the technology so as to know what can realistically be accomplished.
A consulting firm can prove invaluable in the evaluation and selection process for an RPA platform and other related software. But of course, you need to be cautious. A firm may specialize in one platform or even have some type of financial arrangement with the partner, which means you may not be getting the best option. So when interviewing a consulting firm, see if they are vendor agnostic.
Some consulting firms – such as the larger ones – will have their own proprietary frameworks and methodologies. This involves a standard set of strategies that have worked in other industries and that should be more understandable to people who will use the RPA software.
A consulting firm may have created its own software to help with the RPA implementation. Some examples of this include systems to evaluate your company’s processes, prebuilt libraries, and bots.
A consulting firm can be helpful in looking at taking your RPA implementation to the next level. This would be the case with the use of AI, which usually takes much time and resources for an organization to adopt. But a consulting firm can help smooth out the process and get to actionable results faster.
A consulting firm can be a good source of training and should also have educational materials for self-learning. It’s also a plus if there are interactive tools and videos.
A Center of Excellence (CoE), which we’ll cover in Chapter 6, is a team that manages the RPA system. And yes, a consulting firm can help assemble this team and may even be a member of it.
A consulting team should have in-house capabilities to handle the complex issues of compliance and regulations, which can be included within the RPA design. There can also be help with setting up the right structure for governance (risk assessment, oversight, and controls).
The business model may be based on results. That is, there will only be payments if certain milestones are reached, allowing for more alignment of goals.
A consulting firm may provide a “robotics-as-a-service” option. This means that a client can host their RPA system on the firm’s servers, which will likely lower the costs and make it easier for managing the project.
True, there are many RPA consulting firms. This is inevitable as it is not necessarily difficult to start one and the industry is growing at a rapid pace. This definitely adds to the risks. In other words, make sure you take the time to interview a variety of firms and check the references.
It’s also critical to have a firm that is either local or willing to travel to your location. Having a virtual-based RPA consulting arrangment is probably not good enough. A major reason for this is that it will be tough to evaluate your company’s processes.
In Appendix A, there is a list of the top consulting firms that have specialties with RPA.
RPA Consulting: Some Case Studies
For a large organization, an RPA implementation can be a major undertaking. But a consulting firm can definitely be a big help, greatly increasing the odds of success.
Let’s take a look at a case study of Cognizant, which helped a major insurance company with its RPA initiative. Like many others in the industry, the company had a large number of manual processes that could be automated.
At first, the focus was on the operations in Australia with the HR and finance departments. Yet it was a significant endeavor as Cognizant was tasked with assessing the back-office processes of about 2,000 employees. There were extensive conversations with leaders in the departments that helped to narrow the number of areas to target for automation. By doing this, Cognizant identified more than 100 automation opportunities – with the potential for $7 million in net savings.3 This meant that more than 80 people could be redeployed for more value-added work.
However, in order to make the project manageable, Cognizant narrowed the list and then conducted an in-depth analysis of the remaining ones. This also resulted in a detailed report, which set forth the goals, the deliverables, and ROI expectations.
This is not to imply that the insurance company could not have done this on its own. Although, it seems reasonable that the process would have been longer and there would have been more mistakes made.
Genpact, which is a top consulting firm, helped a large aerospace firm with its RPA implementation (it also included the addition of AI). Some advisory services included help with assessing the company’s processes, selecting the software, designing the CoE, and putting together a pilot program. The results? The aerospace company recorded productivity savings of 30% to 50% and the expectation is that there will be an 8X return on investment (ROI) over a three-year period.4
A global bank wanted to find ways to manage higher volumes of transactions without hiring more employees. Keep in mind that the processes were generally complex and the data was scattered across silos. To help with all this, the bank retained Deloitte to provide an analysis of an RPA project. In only six weeks, the first bot was created and deployed successfully. Realizing the quick benefits, the bank aggressively expanded on the effort, with bots for credit card remediation, PDF conversion, and payment processing. The bottom line? The bank had more than 150 bots that processed 120,000 requests per week at only 30% of the costs of using employees. The estimated savings were about $40 million for the first three years.5
A global rental car company wanted to automate its shared services center in Budapest. For the most part, the use of Excel and basic scripting was no longer effective. The firm retained EY for an RPA pilot program, helping with identifying areas to automate, selecting the right tools, and building bots. The potential savings amounted to 10 to 20 full-time equivalent (FTE) employees or €1 million over three years – for a 900% ROI. Because of the success, the car rental firm has looked at extending the project for 200 to 400 FTEs over two years.6
What to Automate?
It seems easy to select what needs to be automated, right? Well, this is true in some cases. But as we’ve seen already in this chapter, there needs to be some analysis of existing processes – to get a holistic view of the workflow. From this, you will be in a much better position to see what needs to be automated.
But it is also critical to understand that RPA is a fit for certain types of automation. It’s not an one-size-fits-all solution.
What’s more, when evaluating what to automate, there should be a group of people with diverse backgrounds – in terms of technology, business expertise and knowledge of the day-to-day operations of a department.
Tedious work: It’s the kind of activity that requires little knowledge. This could be cut-and-paste, and clicking a variety of buttons. Let’s face it, do you want to pay a talented employee do such mundane things?
Time-consuming: Which types of processes result in late outcomes? This could be, say, for creating a report or doing end-of-quarter processing.
Repetitive: The process has a set of steps that rarely change. This is ideal for the mechanical approach of RPA.
Frequency: To get real value, the process is something that is recurring. This may be where a person engages in the activity a few times a week.
Rules-Based: This is because RPA allows you to easily create workflows, which have features like IF/Then logic to carryout actions on a consistent basis.
Clearly Defined Processes: You want to make sure you have some basic metrics, such as the number of steps or a flow chart.
High-Volume: Such activities can be problems for workers. But of course, software is spot-on for handling high-speed processes.
Prone to Error: Are there areas where data input errors are common?
API: If an API is not available, then the process could be a good candidate for RPA automation.
Customization : True, your apps may allow for this. But it is costly. You may have to pay fees for the vendor as well as for hiring third-parties. In other words, the RPA option could be more cost-effective.
Sensitive Data: Access to employees is certainly a risk. But with part of a well-defined RPA bot, you should have much better protection of the data.
Scale: If you do not want to hire more people to ramp up operations – because the current IT setup is fairly rigid – then RPA could be a good option.
Organization: Look for areas within a company where there are persistent silos and bottlenecks.
Yes, this is a lot! And it can be overwhelming. This is why you might want to narrow the criteria down to four to five factors. For example, the common ones used include: tedious work, rules-based, high-volume and repetitive.
Criteria | Score (1-10) |
|---|---|
Frequency (number of times used per week) | 7 |
Potential Time Savings | 2 hours |
Steps in the process | 12 |
Steps that require human intervention | 2 |
Next, you can then compare and contrast the different candidates – and see which ones stand out.
Now as for the first few processes to automate, it’s usually not the best approach to look at cost savings. Rather, the focus should be on looking at areas that will allow for easier automation, such as only looking at a narrow process within a business segment that has a limited number of steps. It’s also helpful if it does not require much approval and help with different departments.
This will accelerate the learning. Essentially, you want to get a quick win – which will motivate the team and allow for more ambitious goals.
Data Migration : Because a company often has a hodgepodge of applications and software, workers will spend much time just transferring information around. It’s definitely a big waste of time and susceptible to mistakes. With RPA, it is relatively easy to handle data migration because of the integrations and low-code capabilities.
Data Updates : What if a customer has a change of address or phone number? How do you update this across different software systems? It can be a pain for workers. But RPA has abilities to create rules to automate this process – and there may not be a need to seek out the assistance from IT.
Tracking: Employees simply do not have the time to monitor the many changes within a department. Because of this, it’s common for problems to slip through the cracks. However, RPA can be configured to detect changes and take action – without human intervention.
Alerts: You can program a bot to detect when there needs to be a decision from a worker. For the most part, RPA is ideal for this.
Rationales for the process
Steps or flow charts
Names of the workers involved
IT assets involved
The requirements for security, compliance and governance
OK, what processes are not good for RPA? According to Aaron Bultman, who is the Director of Product for Nintex RPA: “Tasks involving creative thinking, brainstorming, interacting with the physical world – like pulling papers from a filing cabinet – are better handled by people. But just because a process involved something like this, doesn’t mean you can’t automate the repetitive parts. A workflow automation tool allows you to automate the repetitive stuff and send an assignment to a human worker to complete. Once that task has been completed, the workflow resumes. And in some cases, tasks can be sent off to a RPA bot to be completed on the desktop. You see, when you have both workflow and RPA, you can automate far more in your organization. The idea is to only leave the human work for the human to perform. Nobody likes working on low value, thoughtless or repetitive work.”7
Something else to consider: You might want to start with back-office functions. Why? These tend to be tedious and rules-based. As for front-office applications, there can be much more variability. What might a customer do on the phone? How might there be interaction with a chatbot?
There is also the risk of having a bad customer experience if the bot is not developed properly.
ROI for RPA
Getting a sense of the ROI is tough in the early stages of an RPA implementation. But it’s important to come up with something. This will help to setup some milestones to achieve and also provide a way to track the overall project. But to get a good measurement of the ROI, you will likely need to wait for at least a year.
What is part of the calculation then? The “return” part of the equation is generally the costs that are saved from the RPA project, such as for the less need of full-time equivalent employees (FTEs).
Example: Suppose you want to automate the invoicing process and it requires four FTEs who are paid $20 per hour. They work on the process 40 hours a week. Thus, the total labor costs will come to $166,400 per year.
As for the RPA system, it has reduced the costs by 60% or to $58,240 per year – which puts the net gain at $108,160 ($166,400 minus $58,240).
Year | Amount Saved | ROI |
|---|---|---|
1 | $108,160 | 35.2% |
2 | $216,320 | 170% |
3 | $324,480 | 306% |
In this example, this project would certainly be a big winner.
Accuracy: What has been the decline in errors compared to handling the process in a manual way? Also, is the data quality better because there are fewer mistakes with data input?
Customer Satisfaction: Has there been an improvement? One common metric to look at is the Customer Satisfaction Score (CSAT), which is a customer survey that asks “How satisfied were you with your experience?” Or you could use the Net Promoter Score (NPS). This shows a customer’s interest in recommending a company’s product or service – and the index ranges from –100 to 100.
Agility: How fast is the process now? After all, a bot can work at high-speed on a 24/7 basis.
Employee Satisfaction: You can send out a periodic survey to see if there has been an improvement. Furthermore, look to see if absenteeism and turnover rates have dropped.
Innovation: With the RPA handling mundane tasks, has there been more time for employees to provide higher added value? Are they creating new approaches or innovations?
Analytics: You can get better tracking of your processes, which can provide even more insights on what areas to improve upon.
You can note some of the qualitative benefits. Has there been more reliability to the process? Is compliance improved overall? With such things, you can include them in a report, which should help with the buy-in for the RPA project.
As for costs, the main line item, of course, is the license or subscription fee for the software. There may also be other IT expenses, such as for purchasing prebuilt bots, servers, hosting services, and other software. Then you will need to account for the labor costs for the implementation, bot development, and monitoring, which may include the retaining of a consulting firm. This is why many companies and consultants have used a more comprehensive measure called total cost of ownership, which looks at indirect costs over the life cycle of the automation as well as direct costs (i.e., software licenses).
The good news is that – over time – there should be economies of scale. In other words, certain line items , such as for training, may decline. But then again, there will also be declines with the return because there will be fewer opportunities for “low hanging fruit” for automation.
RPA Use Cases
As RPA has been around for a while, there are many use cases for different departments and industries. This can certainly provide ideas on what areas to start an implementation. Might as well focus on those places that have worked for others, right?
Definitely.
Customer Service | Validation of checks Customer reminders and notifications Processing of customer feedback Out-of-hours responses |
Finance | Credit approval Collections process Statement reconciliation Invoice creation Budgeting Accounts payable Accounts receivable ERP data entry Profit and Loss preparation |
HR | Talent recruitment Employee onboarding Termination Payroll Benefits administration Employee training Compliance reporting Employment history verification Expense management and reimbursement Time record validation HR spend analytics Requisition management |
IT | Configuration and installation Server monitoring Batch processing Email processing and management File management Backups System queries Data migration Password reset and unlock Data validation Security monitoring Event management Application testing Help and service desk management |
Marketing | List/contact management Social media tracking Email campaigns CRM updates |
Procurement | Management of work orders Processing of returns Vendor administration Freight management Contract management |
Tax | Item preparation Posting of accounting items Entity management Modelling |
Banking | Loan processing Account opening Account closure Transaction processing Risk management Approvals Confirmations Fraud detection |
Insurance | Claim verification Appeals Commission calculations New customer packages Policy changes |
Telecom | Customer dispute resolution Phone number porting Credit checks Processing customer requests Collections |
Retail | Call center Demand forecast Returns Promotions/discounts Supply chain management |
Healthcare | Patient management and administration Appointments Insurance coding Patient data migration |
The Plan
So far, we have looked at the ways to understand and prepare for an RPA implementation. As you hone in on the key areas for the project, it will be time to put together a written plan. It does not necessarily have to be long or detailed. Instead, the focus should be on developing a clear and effective road map.
Reduce the FTE time worked on processes
Improve cycle and turnaround times
Lower the number of errors
Again, estimating these will, in most cases, be far from accurate. But it is important to set forth clear-cut milestones to hit. And if they prove too onerous or not particularly challenging, you can then make midcourse adjustments.
The RPA plan should also be consistent with the digital transformation goals of the company. This should also be backed up with a champion or sponsor of the project, hopefully from the executive ranks. This will be an indication of the importance and commitment to the RPA project.
Next, the plan needs to set forth the members of the core team. Often this is a diverse group, including people from different departments. There will also need to be a strong leader since the team members will likely not be working on the RPA project on a full-time basis.
No doubt, the plan must be clear-cut on the roles and tasks for each team member (in chapter 6, we’ll look at how to form a CoE, which can serve as the team). And then it’s a good idea to involve the IT department in the process so as to make sure that there are no issues with integration, security, and governance.
Then the plan can look at the factors for the RPA vendor evaluation. In Chapter 5, we’ll look at this in much more detail.
Finally, the plan should have a project road map that shows timelines and deliverables. Usually, this will begin with the pilot or proof of concept. As noted earlier in the chapter, the first project will be more limited, which should allow for more learning and getting to faster results. But the plan can also cover the longer range goals for the RPA efforts, in terms of areas to automate and departments to cover.
Conclusion
Even before you create your first bot, there is much to do! But these steps will be essential for success. If an RPA project fails in the early stages, it could ultimately doom any chances for the use of the technology. Management may simply just cancel the program.
Then when can you start creating bots? Well, not yet. In the next chapter, we’ll take a look at how to evaluate an RPA vendor.
Key Takeaways
While RPA is an easy setup, this does not mean you should quickly start creating a bot. Instead, a better approach is to do some planning. In the first step, you can have brainstorming sessions or even put together a workshop. It’s also important to provide some initial training about the fundamentals of RPA.
To help with the planning and implementation of an RPA system, many companies do seek the help of a consulting firm. Even though there are upfront costs, the benefits can be significant. A quality consulting firm will have full-time specialists who have seen many RPA systems. There could also be help with other areas like assessing company processes, implementing AI, and even hosting the software. Some firms have created their own technology frameworks and bots, which can help streamline the process. Finally, a consulting firm will have the experience with multi-RPA implementations. This will help with better planning and the avoidance of certain issues.
Spend time in considering what processes to automate. The fact is that RPA is fairly limited in terms of the use cases. For example, it is really for those processes that are tedious, repetitive, rules-based, and prone to error. It’s also good if they are recurring in nature (maybe involving an employee several times a week). On the other hand, if the process requires much judgement, ingenuity, and creativity, then RPA is not a good option.
When looking at the first processes to automate, it’s usually better to look at those that are relatively easy and have fewer steps. The reason is that you want to learn from the experience. Over time, as you get better at creating automations, you can take on more complex processes.
To evaluate which processes to automate, you can use a diagnostic. You may want to look at only a handful of factors, such as those that involve tedious work, are rules-based, have high volume, and are repetitive. You can then score each and see which ones stand out.
Some of the tasks that RPA can help automate regardless of the industry include data migration, data updates, and tracking.
Even early in the project, it is advisable to estimate the ROI. This helps with providing clarity of the objectives and should help with buy-in.
The “return” in ROI for an RPA project is generally the cost savings, such as in reduced number of FTE hours. But there are other benefits like fewer errors, improved compliance, and higher customer satisfaction.
The “investment” part of the ROI involves the costs of the software, the employee time, consulting fees, and other software/hardware costs.
The written plan for the RPA implementation does not have to be long. Instead, think of it as a road map. In the plan, you should provide details about the objectives, roles, and deliverables (with timelines). It is also important to have a champion, from the executive ranks, who will make sure the project remains a priority.