great place to work

The Ultimate Guide to Implementing RPA Successfully in 2018

The Ultimate Guide to Implementing RPA Successfully in 2018

Executive Summary

Do you know what gets executives really excited? Automation and robots. Do you know what scares the bejesus out of everyone else? Automation and robots.

That’s why it continues to be a hot subject as more companies move to implement Robotic Process Automation (RPA). It almost works too well.

RPA isn’t a new concept by any means. It’s been around for a long time and is starting to gain momentum in large corporations as they’ve seen the results of cost savings from previous RPA implementations.

In this article, we break down what RPA is and explain how you can use it for your organization as well as how to best implement it – from creating the business case, to the initial discovery workshops and implementation, to testing, and rollout.

The Ultimate Guide to Implementing RPA Successfully in 2018

What is Robotic Process Automation (RPA) and why would I use it?

RPA does exactly what it promises: It automates processes via robots. Well, more like digital robots AKA, “bots.” There are, unfortunately, no real robots involved in RPA. RPA tools mimic the same manual processes performed by a human in a digital environment.

Here is an example of a small manual process that can be automated via RPA:

1. Cindy receives an email notification that an insurance claim has been filed.

2. Cindy opens up the email notification, copy and pastes the contents of the email into another system (system #1), and presses submit.

3. Cindy waits for the confirmation page to load, and after it does she copies the unique ID generated by the system and pastes it into another separate, unconnected application (system #2), then she presses submit.

4. Repeat.

In this instance, a RPA bot would automate that entire process because it is defined and contains digital data from the beginning. The benefit of RPA in this case is that the systems (email, system #1, and system #2) don’t need to be configured to speak to each other. RPA is literally mimicking Cindy’s mouse movements to perform the actions. This is one small use case of RPA. Find below some more examples in graphical format.

Key areas where RPA can automate the process

We can break it down even further with more examples of lead generation and operations.

Break down – Some of the top use cases

The evolution of digitization and how RPA is part of it

In the following image you can see where we are in the evolution of digitization. ERP’s and eCommerce sites are in place, and through the magic of mobile, they are putting a lot of power in consumers’ hands.

The problem with this natural evolution is that many of these new systems can’t talk to each other. So manual work by humans is needed to keep the process moving and get information from one system into another. To be more cost efficient, companies are looking at ways to automate these processes without initiating complex and expensive IT projects to connect the systems. This is where RPA comes in. It’s a way to get data from one system into another.

A big benefit for any company in a highly regulated industry

Banks, insurers, pension funds and other financial services institutions operate in a highly regulated industry and are faced with high demands for auditability, security, data quality, and operational resilience. Robotic process automation allows modern banks to meet these demands and achieve significant operational efficiency.

Common myths about RPA: Here’s what it can and can’t do

You know that video from Boston Dynamics where the robot opens doors and does backflips? That’s a real robot. It also has the capabilities to think for itself and walk around barriers. It doesn’t need structure. The word robotics is the first misconception about RPA, especially because you can use RPA with machine learning and artificial intelligence.

Myth #1 – It works well on unstructured data

The Boston Robotics robot is smart enough to evaluate its surroundings and make an appropriate decision based on the goal and the data it has collected. It can walk around moving blocks, stop when someone is coming and open doors it has never seen before.

RPA? Not so much.

RPA works when the data is structured and has a high transaction volume, and the decision paths are fairly standard and have direct digital access to the data.

Myth #2 – Automation will help improve our bad processes

The way to really think about RPA is that the “robots” will do exactly what the humans used to do. If the humans were inefficient because of unstructured decision making, then the robots are almost guaranteed to be worse than the humans. RPA works great when the processes are already defined and easily automatable.

Here is a simple broken down process of how to determine which process can be automated.

Process A:

1. AP receives 100 invoices a day.

2. AP Person 1 categorizes the invoices based on data from the invoices and clear instructions on how to categorize them.

3. Invoice gets routed to right team for approval.

4. Team member gets notification and approves invoice.

5. Payment is made upon approval.

Great! This can be automated

The AP Person has a clear process on how to categorize invoices based on data characteristics. This makes the job of for RPA easier because the data is structured and clearly defined.

Process B:

1. AP receives 100 invoices a day.

2. AP person prints out 100 invoices during weekly invoice meeting with multiple team members and each team decides what invoices need to be approved or rejected.

This can’t be automated with RPA! But this situation can be entirely avoided with RPA.

The process is broken. How to categorize invoices isn’t defined and there is no clear data structure for making decisions. Too much human judgement and other contingencies involved.

The Ultimate Guide to Implementing RPA Successfully in 2018

Pre-work part-I: Is RPA worth the cost of investment? Here’s how to calculate the ROI of RPA

There are three reasons why companies choose to implement RPA:

1. It significantly boosts productivity with minimal process change.

2. It’s easy to calculate the ROI.

3. It’s a fresh alternative to the large complex projects of typical BPM programs.

See the image below for more details on each reason.

1.1 Start with a survey of multiple business functions

Do business leads feel that certain manual activities can be automated? If so, how many?

1.2 Calculate the REAL ROI

It’s easy to present a powerpoint slide about how RPA will save the company $200,000 a year if implemented, which can be entirely true. What’s usually missing from these presentations is all of the related effects of RPA which can actually increase costs that aren’t easily measured. For example, an automated process can trigger more support calls and complaints from other parts of the business where RPA isn’t working properly. Also, if it reduces headcount, are you really saving on headcount or are you transferring that resource, and also cost, to another department? And don’t forget about implementation costs. All of a sudden your $200,000 per year savings doesn’t look as nice. So, calculate the real value of an RPA and calculate the external factors.

1.3 Calculate the opportunity that RPA can make in the next 6-12 months

Let’s say that it makes financial sense to initiate RPA for certain identified processes. The next step is to visualize what other opportunities will occur in the next 6-12 months.

Pre-work part-II: Organizational preparation

Due to the business impact of RPA, this step shouldn’t be taken lightly. We have seen great RPA projects fail mainly because the organization wasn’t prepared for the impact, which caused a lot of issues downstream. The goal of this section is to be ahead of the game. Making sure the right people within the organization are aware of its potential impact and are actively engaged in the project is vitally important.

2.1 Create a governance board with IT and business functions. Communicate the need for RPA and the potential impact once it’s implemented.

Governance is not a new word and is business-as-usual for most corporations. Creating a governance board should be one of the main priorities for the sponsor of the project. The executive sponsors on the governance board should include the business executives, technology executives and relevant resources of divisions that will be impacted by this change. That could also mean including HR on the governance board to control the message.

2.2 Document which high level processes can be automated.

The workshop will be used to flesh these high-level processes out, but it’s always a great start to get the business and IT teams to think about what processes can be automated. This is the responsibility of all executive sponsors and members of the governance board. Collect all of those ideas in one place.

The Ultimate Guide to Implementing RPA Successfully in 2018

Discover part-I: Workshop and roadmapping - The first week of the project engagement

So you’ve decided to move ahead with your RPA project. The most critical part of an RPA engagement is the first week of the discovery process. Now, let’s walk you through how to best structure your future RPA engagements for maximum success. These guidelines are based on our years of experience implementing RPA for many organizations.

The first thing to be done is getting your IT, Business, Governance and most importantly, company politics aligned. The word “robots” and “automation” used in the same sentence will easily trigger the fastest 100 email thread in the history of your company. So this needs to be taken very seriously, and communication regarding this should be tight. When we start a RPA project, our next step is an in-person workshop to help flush out what’s being implemented.

3.1 Invite all the decision makers impacted by RPA (executives & sponsors)

Depending on the complexity and number of processes, the workshopping could last anywhere from 4 hours to 4 days. One of the first things we do is schedule a workshop with all the decision makers in one room. Every owner of the business and relevant IT Directors should be part of this workshop. This is also not an activity that can be done over the phone. It’s interactive and requires a lot of communication.

Since this meeting involves a lot of people, it needs to be organized and fluid.

Here are the main deliverables for a workshop:

All day workshops tend to be exhausting and oftentimes, things will be missed. By sending out a clear definition of everything that was decided on, it gives the teams a few more days to fill in any gaps. Since this meeting involves a lot of people, it needs to be organized and fluid.

Here are the main deliverables for a workshop:

All day workshops tend to be exhausting and oftentimes, things will be missed. By sending out a clear definition of everything that was decided on, it gives the teams a few more days to fill in any gaps.

3.2 Structure the workshop for maximum effectiveness

Since this meeting involves a lot of people, it needs to be organized and fluid.

Here are the main deliverables for a workshop:

1. A defined list of processes that can be automated and ranked by priority

2. Detailed process flows for the main processes to be automated.

3. Strategy roadmap and Return On Investment calculation

4. Project development plan

5. Governance model and dependencies

3.3 Send out results of the workshop and clearly communicate next steps

1. A defined list of processes that can be automated and ranked by priority

2. Detailed process flows for the main processes to be automated.

3. Strategy roadmap and Return On Investment calculation

4. Project development plan

5. Governance model and dependencies

3.3 Send out results of the workshop and clearly communicate next steps

All day workshops tend to be exhausting and oftentimes, things will be missed. By sending out a clear definition of everything that was decided on, it gives the teams a few more days to fill in any gaps.

Discover part-II: Design the architecture and choose the tool(s)

The biggest misconception about RPA is that it’s one-size-fits-all, when in reality it can be a variety of tools that do widely different things under the guise of RPA. In this section, we’ll overview the different types of tools you can use for RPA. In the workshop and discovery part of the project, all of the requirements were listed, so the next step is to define the right tools.

4.1 Choose the right tool

Standard RPA tools – Many of the best RPA tools are simply powerful macro bots that can mimic human behavior. As long as there is a standard process involving digital data, there are some great tools.

Here are our top recommendations of RPA tools:

• Uipath

• Blue Prism

Chatbots – A good example of a use case for Chatbots is when a human used to ask the same set of questions to a customer. Instead of having a human asking and recording the answers to those questions, a Chatbot can ask those questions and even ask follow up questions, depending on the answer to the previous questions.

Chatbots are already replacing contact centers in responding to the FAQ-type questions. They are also being used within enterprise. The use case can range from letting you know the status of a tech support issue, to answering scientific questions on pharmaceutical formulas or investment models.

A chatbot can be script driven, data driven, or inference driven. Script driven chatbots are for basic purposes such as providing application status. Data driven chatbots can be used to support complex scenarios, such as the eligibility for a credit card. Inference driven chatbots can be used in complex, unstructured, and unknown scenarios. For instance, a robo-advisor in a volatile market will make a recommendation while considering the investor’s sensitivity to risk.

As chatbots get smarter, they will become more popular.

API Integrations – Integrating with API’s is oftentimes a last resort when it comes to RPA, considering the whole purpose of RPA is to mimic screen movements. However, if strategically done, API integrations can help expedite a RPA that can’t be done with a standard screen-mimicking RPA tool.

How to execute?

At this point, an RPA implementation runs similar to any technology implementation. There is no golden answer to how you implement it. It’s all based on your company’s and implementation vendor’s preference.

5.1 Development based – Agile or Waterfall methodology?

There isn’t one clear answer to this. In some instances it makes sense to do the easiest process first via an Agile methodology and push it to production. In this case, you can get it running as soon as possible, work out the kinks, and use it as a baseline for future implementations. This also helps you build more confidence with other team members that implementing RPA is a benefit to the company. Both are development oriented. RPA involves more business intervention and continuous testing and approval. DevOps is highly preferred while implementing RPA.

5.2 Transaction based – Pay as you go

If you are partnering with a technology vendor, a transaction based model makes more sense. When your transactions increase, the overhead also increases. With RPA, the overhead reduces. You pay only for the transactions.

5.3 What happens if the bot stops working? Define a risk assessment and business continuity model

What if the invoicing system gets upgraded and causes the data mapping to stop working?

• Does the AP team stop approving invoices?

• Do you wait for IT to fix it?

• What’s the process for handling emergency invoices?

It’s fine to trust the robots. But have a plan just in case the robots get stuck. Create a risk assessment and define any points of failure, and have a backup plan. Assume anything that can go wrong, will go wrong. Alternatively, work with a technology vendor and opt for a managed services model. A managed services model is where the vendor will take care of the entire process and is responsible for the uptime and process quality.

Go-live & support

The three things we need to have before we go-live:

• All signed-off test documents

• Sign-off from all executive sponsors

• Sign-off that all backup procedures are ready to go-live if a “disaster” occurs which causes a RPA procedure to stop working

1. A dedicated control center with all the relevant implementation members and vendor support. We recommend having this until the RPA is stabilized. The control center should have:

2. Go-live checklist updated frequently

How you go-live depends on the methodology that you’ve chosen. For RPA, since it involves automating an entire process, it is important that close attention is being paid to the individual transactions as they go through the system. Here are the things we recommend to have during go-live:

• Implementation members

• Team members who were responsible for manual activities

• Tech representatives of the vendor available in case software expertise is needed

• An RPA checklist is critical not only for ensuring things are going smoothly, but also for reporting to the executive committee

The Ultimate Guide to Implementing RPA Successfully in 2018

Bonus: How to choose a RPA technology service provider?

There are two ways to implement RPA. You can do it straight through the vendor or you can hire a technology consulting firm to do it for you. What we’ve seen from our clients is they will often get a proof-of-concept from a tool vendor, and then after they see it’s a fit for their company they will look for a vendor. We’re a little biased since we specialize in RPA; so here is what to look for in vendors:

1. Do they have dedicated resources to RPA?

Just because they have it on their website as a side note, doesn’t mean they have resources dedicated to RPA. It’s important that RPA is an actual practice within the firm since there are a lot of various  ways to implement it.

2. Do they have case studies of work done in your industry?

We have seen the biggest impact of RPA in financial services and mainly in Insurance. We’ve been able to build up extensive knowledge in business processes and tools involved to spot issues before they arrive.

Here is a link to our RPA case study

3. Do they pitch a tool before they know your requirements? If they do, be cautious.

Some IT consultancies have agreements with vendors that unfairly bias the tools they select during the implementation. The process and roadmap defines the tools you use, not the other way around. Do your own due diligence on tools and look into the full cost, including the licenses of each robot.

Case study : How we reduced a claims process from 14 days to 2-3 days using RPA

An average claim takes 10-14 days to process if only humans and ERP systems are involved. Here’s what a sample manual process looks like for claims processing.

PHASE 1: Semi automated: Bots and humans co-work to process claims

In the next phase, instead of jumping straight from manual to a 100% RPA and machine learning solution, we decided to prove the value of RPA with automation and human involvement. It was a logical step to continue.

PHASE 2: Automated: Bots do the hard work

In this phase, we took it to the next level and automated more of the complex processes and eliminated most of the human interaction.

After all was said and done, we reduced the claims processing period by 97%, which was a huge win for the client. If you want to learn more detail about this case study, download it here.

Radhakrishnan K R Practice Head – FinTech RapidValue Solutions

Sivagnanam V Practice Manager– FinTech RapidValue Solutions

If you’re looking to implement RPA, this guide has all the details you need to be best prepared for it. If you would like RapidValue’s expertise in a RPA implementation, please reach out to and we’ll be in touch with you shortly.

How can we help you?
Please share your details to continue to read the whitepaper.