AIMultiple ResearchAIMultiple Research

21 RPA Pitfalls & Audit Checklist to Tackle Them in 2024

21 RPA Pitfalls & Audit Checklist to Tackle Them in 202421 RPA Pitfalls & Audit Checklist to Tackle Them in 2024

Robotic Process Automation (RPA) is gaining acceptance, especially in the finance and telecom industries however we have spoken to several executives and there are plenty of RPA failure stories, too. A recent survey showed that >40% of RPA projects fail to deliver expectations in terms of

  • implementation time
  • implementation cost
  • cost savings due to RPA
  • benefits to analytics

We outlined some of the most common pitfalls that lead to these gaps between reality and expectations. We explain these points in detail below:

  • Organizational pitfalls:
    • 1- Lack of time commitment from local team
    • 2- Lack of leadership buy-in
    • 3- Lack of IT support
    • 4- Lack of support from Analytics/Data function
    • 5- Lack of support from HR
    • 6- Unclear responsibilities
    • 7- Companies lack a clear RPA strategy
  • Process pitfalls:
    • 8- Choosing a process that changes frequently
    • 9- Choosing a process with an insignificant business impact
    • 10- Choosing a process where errors are disproportionately costly
    • 11- Choosing a process that involves higher level cognitive tasks
    • 12- Choosing a complex process. Though its sub-processes are simple, process itself may be complex if it has too many sub-processes
    • 13- Choosing a process where better custom solutions exist
    • 14- Striving for end-to-end automation when it is not cost-effective
  • Implementation pitfalls:
    • 15- Pursuing in-house RPA development with in-house teams that do not have enough capacity
  • Technical pitfalls:
    • 16- Choosing a solution that requires intensive programming
    • 17- Not relying on RPA marketplaces and other readily available tools
    • 18- Choosing a solution that did not demonstrate scalability
  • Post-implementation pitfalls:
    • 19- Not building for scalability
    • 20- Not taking maintenance needs into account
    • 21- Not securing RPA privileged credentials

Organizational pitfalls: Alignment is key for any project’s success

Especially in projects where there’s no outside implementation partner, organizational alignment is key because your organization will be responsible for the whole solution. Both local team and leadership needs to be fully on-board, with top management regularly reviewing progress and local teams devoting significant time to automating processes getting help from departments like Strategy.

Adjacent teams that rely on automated processes also need to be notified and convinced in advance and especially in the beginning of automation they should be on the lookout for any issues.

These are not only implementation related issues. Once the RPA solution is rolled-out, it will require maintenance as processes are changed to make them more efficient, effective or compliant with new regulation. Satisfying these maintenance needs are important and can be challenging if companies do not dedicate enough resources and management attention or if they do not clarify responsibilities.

Do not forget to get support from these key functions

IT

IT roadmap needs to be examined before choosing the RPA solution. For example, if IT is planning to migrate to Citrix, that will have implications for the RPA tool that was chosen

Additionally, IT can act as a coordinator in these tech purchase decision by other units. If there’s already an RPA implementation in the enterprise, IT should bring both things together and help them learn from one another’s experiences. Shadow IT which leads to different divisions using a myriad of tools leads to sub-optimal IT costs and data silos.

Data/Analytics

Data/analytics is on most senior leaders’ agendas and bots have the potential to create significant amounts of data. If analytics function is involved early on, the format, frequency and other important decisions about the data created by bots can be considered early on. This leads to bots creating valuable data as opposed to simple diagnostic information which is the case in most bot installations. However, benefits of RPA to analytics should not be too exaggerated as we explained before.

HR

HR is important to get aligned with or else RPA training programs may never take their place in the corporate training schedule. RPA training is important to reduce reliance on RPA consultants and empower employees.

Company having a clear RPA strategy is key for sustainable RPA deployments

There are various models of RPA deployment or maintenance. It is important to decide the company’s RPA approach to ensure that teams do not waste time creating an RPA approach from start and end up creating redundant responsibilities. As PwC report points out, an RPA Center of Excellence (CoE), IT, finance, or the teams in charge of the process could be responsible for the RPA deployment. Additionally, company could be relying on external consultants for RPA deployments.

Different organizations give RPA responsibility to different units
Source: PwC

Process pitfalls: The most important decision is the process to be automated

It is critical to fully understand processes before choosing the ones to automate. PwC claims and case studies show that conducting a RPA pilot project often takes 4-6 months instead of the expected 4-6 weeks because businesses don’t have enough knowledge about their as-is processes.

Ideal process is impactful, simple, does not require high level cognitive tasks, lacks a custom solution and is difficult to be automated with non-RPA techniques. Let’s explain all these points:

Business impact is key to excite the organization. RPA project on a process with low business impact will have little momentum. Processes with high business impact tend to be high volume, high effort processes that touch the customer. Nothing like telling a CEO we can approve loans in 2 minutes rather than 2 days.

Process should either be fault tolerant or these needs to be a system for quality assurance. Since RPA bots rely on UX to complete tasks, they can make errors when there are UX changes. In highly critical tasks, RPA may not be the best choice. However, RPA can be deployed for almost all processes as long as the cases that lead to costly errors are verified via other mechanisms (potentially including manual controls) as well.

Process should not rely on ill-defined, high-level cognitive tasks. Reading an email that explains a number of tasks which include communicating with a client and reviewing an advertising image are quite simple tasks for a marketing professional. However, currently these are not well defined tasks and are therefore not suited for automation. For example, it’s difficult to explain what makes a good advertising image. This does not mean that such automation programs are impossible. An automated system could use crowd-sourcing to pick right advertising but it would be costly, slow and hard to program which are not the right qualities in good automation software projects.

Though some high-level cognitive tasks are hard to automate, some tasks that require significant cognitive capacity are being automated. It is best to look at cognitive requirements on a process level. For example getting data from documents and processing that data requires significant cognitive capabilities. However, invoice capture is possible at high accuracy with deep learning. RPA tools can be combined with deep learning plugins to automate that process.

Process complexity is a separate issue from high-level cognitive tasks. A process can involve only low level cognitive tasks like adding numbers, copy pasting and so on. However, based on different input, different set of instructions may need to be executed. For example, depending on a user’s answer to a question, a different department may need to handle the user’s request through a different process. Such a process can be complex and it would be difficult to extract the correct process flows in different scenarios.

Currently it requires a lot of manual process data extraction, interviews and lengthy pilots to successfully automate such processes. However, this is an ongoing area of research for RPA vendors and startups which aim to auto extract process data from logs and videos to successfully automate complex processes. We call these self-learning automation solutions. Newer solutions called cognitive automation or intelligent automation (depending on the company promoting the solution) are able to watch as automatable work is performed by humans, learn the automation needed and takeover when ready. We are investigating such innovative solutions and listing them as they become available.

Custom solutions tend to outperform generic solutions and RPA is a very generic solution. For example, Anant Kale from AppZen recently mentioned how some companies are trying to use RPA for expense audits. I think that’s a great example for a process where a high quality custom solution exists. No RPA solution will be able to go into as much depth as AppZen can over an expense item.

AppZen has a database of fraud patterns and built a knowledge layer of where and how people spend for travel and entertainment (T&E). An RPA solution will not have access to any of that data and will likely perform poorly compared to a custom expense auditing solution. However, managing multiple solutions brings new IT complexities so better make sure that the new custom solution is worth the effort of migration.

RPA is not the only mode of automation. Replacing legacy systems or building powerful API interfaces to legacy systems can help you automate numerous processes with less effort than building RPA solutions. Because RPA systems use imperfect screen scraping solutions, upgrading legacy systems offer faster and more accurate automation solutions.

See our in-depth article about identifying and prioritizing processes to automate with RPA for more details.

Once RPA demonstrates its value, keeping the organization focused is hard

The first process to be automated will likely be selected with a robust process. If that automation delivers significant value, all senior managers will be excited to jump aboard and start automating their processes. This can lead to a loss of focus as RPA experts are stretched thin due to demands from varying departments. Automation of less critical processes will deliver less value than initial pilot and this can lead to an “automation-fatigue”. Though numerous departments spent significant effort to automate processes, they will end up with little improvement.

To keep the organization motivated, RPA experts should be focused on a limited number of high impact projects. As RPA expertise increases in the organization, individual teams will start to take initiative and automate their own processes. This is the ideal state as automation will improve operations without the need for significant time commitment from senior leadership.

Full process automation is desirable but it may not be economical

Many processes are 70-80% automatable without great difficulty. However, as the level of automation increases, businesses face diminishing returns. Automating a process completely may be five times more expensive than automating a process up to 80% because that additional 20% will require automation code that’s a lot more complicated than the code required to automate up to 80%. Process redesign, keeping human in the loop for edge cases are all solutions to operate 80% automated processes with optimal efficiency.

Implementation pitfalls: RPA development is a focused effort, rely on teams that have capacity to deliver

Deploying RPA bots require understanding processes and programming RPA bots. While these take weeks, they require focus. Unless there are teams within the organization who have the time for RPA deployment, it would be wise delay the project or rely on consultants to complete the RPA implementation.

Technical pitfalls: RPA is an evolving field, don’t buy outdated solutions and leverage the full capacity of your chosen solution

Especially when you are outsourcing your RPA setup to a consultant or BPO, keep in mind that they may have a conflict of interest. For example, programmable solutions take longer to implement and therefore result in longer billable hours, however programming time may be reduced with low code/no code solutions.

Bankers, especially those on the technical side, like to boast how their banks are actually tech companies and how they are using the state of the art. However when we start discussing how they implement RPA solutions, some have not even heard about self-learning or low code/no code solutions. We discussed self-learning solutions above, another interesting new area is no code RPA solutions. While normal RPA solutions require intensive programming, no code RPA solutions replace time consuming coding with recording and drag and drop interfaces which aim to democratize RPA.

Reusable RPA plugins/bots which are available on RPA marketplaces, reduce RPA development time and save your teams from reinventing the wheel. Ensure that your teams make full use of the RPA tools that your chosen RPA platform offers. Most leading RPA companies have RPA marketplaces with reusable code. Feel free to read our RPA marketplace or reusable RPA bots articles to learn more.

Finally, using bots that have been proven in large deployments reduces future risk of scalability issues. Most major RPA providers have large (100+ bots in a company) deployments so this should be a smaller concern. However, it would be good to check your RPA software providers’ largest deployment size.

Post-implementation pitfalls: These can slow down or even stop an organization’s transformation

Scalability

Scalability is widely quoted as a major issue especially for Fortune 500 organizations looking to scale up their RPA implementations. Managing an RPA installation involves starting and stopping bots due to business necessities, managing the maintenance process, ensuring that error rate is acceptable. RPA management should also demand very low time commitment to ensure that RPA’s benefits are optimized.

Complexity of managing an RPA installation can grow quickly as number of bots, issues encountered by bots and processes impacted by bots grow. Ensuring that bots are audited post implementation, simplifying bot architectures and following a gradual approach to automation can help facilitate management of RPA installations.

Given increasingly large RPA installations, it seems that vendors are effectively tackling this problem. For example, UiPath in partnership with IBM, Accenture, EY and PwC rolled out RPA bots in Sumitomo Mitsui Financial Group to automate activities in 200 processes leading to 400K hours of annualized savings. 400K hours/year is approximately worth 250 FTEs making this one of the largest RPA deployments globally.

Maintenance

Maintenance is the most important post-implementation challenge. Changes in the regulatory or business environment will sooner or later require changes to bots. Since most bots are programmed, following software best practices in programming allows bots to be maintained with relative ease. Still, changes need to be prioritized and necessary effort needs to be devoted to bot maintenance.

RPA essentially adds a new responsibility to process owners. While they will likely be managing a smaller workforce that produces higher quality results, they will need to allocate time to manage and maintain their bots.

Security

Deployment of RPA technology also means another touchpoint businesses need to secure. RPA bots require privileged access to log in to ERP, CRM and other business systems to extract and move data through a process from one step to the next. Since RPA software interacts directly with your organization’s business systems and applications, it may introduce significant risks such as

  • hard-coding privileged credentials directly into the script or rules-based process
  • retrieve the credentials from an insecure location, such as off-the-shelf solution configuration file or database.

See our article on how to measure RPA success post implementation.

Other surveys of RPA customers also show companies struggling with these pitfalls

Forrester

Recently the analyst firm Forrester surveyed a group of RPA customers as part of its research for UiPath and published its results as a report. We have seen similar issues highlighted

List of issues faced by of RPA customers
Source: Forrester

We do not have access to the free text responses but the first issue definitely highlights the implementation problems these firms are facing. Scaling RPA solutions require effort and companies are probably having difficult time dedicating personnel to automation efforts while they need to ensure the smooth functioning of their business. And scaling automation with consultants may not be economically viable for processes that are not among the most valuable and frequent processes of the company.

Second issue highlights challenges of maintaning bots while processes change. Clarifying responsibilities about maintenance is the first step in tackling this issue.

The last 3 issues indicate customers’ lack of satisfaction with the RPA solution they chose. Choosing the best available product can help minimize these issues. Our content below can help you choose the right RPA product for your business:

PwC

PwC’s survey also highlighted similar issues

Source: PwC

So far these have been the cases I frequently heard from the trenches. Will keep on updating as we hear more implementation stories. Please leave a comment if you have other stories.

Now that you know RPA pitfalls, you could benefit from our popular step-by-step guide to RPA implementation.

And if you believe your business will benefit from an RPA solution, feel free to explore our data-driven hub of RPA and automation vendors.

And let us guide you through the process:

Find the Right Vendors

Sources:

Featured image

Access Cem's 2 decades of B2B tech experience as a tech consultant, enterprise leader, startup entrepreneur & industry analyst. Leverage insights informing top Fortune 500 every month.
Cem Dilmegani
Principal Analyst
Follow on

Cem Dilmegani
Principal Analyst

Cem has been the principal analyst at AIMultiple since 2017. AIMultiple informs hundreds of thousands of businesses (as per similarWeb) including 60% of Fortune 500 every month.

Cem's work has been cited by leading global publications including Business Insider, Forbes, Washington Post, global firms like Deloitte, HPE, NGOs like World Economic Forum and supranational organizations like European Commission. You can see more reputable companies and media that referenced AIMultiple.

Throughout his career, Cem served as a tech consultant, tech buyer and tech entrepreneur. He advised businesses on their enterprise software, automation, cloud, AI / ML and other technology related decisions at McKinsey & Company and Altman Solon for more than a decade. He also published a McKinsey report on digitalization.

He led technology strategy and procurement of a telco while reporting to the CEO. He has also led commercial growth of deep tech company Hypatos that reached a 7 digit annual recurring revenue and a 9 digit valuation from 0 within 2 years. Cem's work in Hypatos was covered by leading technology publications like TechCrunch and Business Insider.

Cem regularly speaks at international technology conferences. He graduated from Bogazici University as a computer engineer and holds an MBA from Columbia Business School.

To stay up-to-date on B2B tech & accelerate your enterprise:

Follow on

Next to Read

Comments

Your email address will not be published. All fields are required.

13 Comments
Purnima Shukla
Jun 21, 2021 at 04:40

RPA, AI, ML, etc are future skill courses that are becoming in demand and popular among the students. This blog properly explains the challenges faced by the world of RPA. Would love to read more on this.
Hope to revisit soon!

Sonika Aggarwal
Nov 25, 2020 at 06:41

The majority of the process pitfalls in RPA can be avoided if the enterprise uses a process discovery tool like Skan.ai to analyze the process and create a holistic picture. It can help identify the perfect process candidate, analyze to what extent it can be automated, and help in continuous monitoring for improvement.

Cem Dilmegani
Nov 28, 2020 at 19:52

Hi Sonika, thank you for contributing,
We hear that from numerous process mining/discovery vendors, what makes Skan.ai different?

Sonika Aggarwal
Dec 16, 2020 at 12:09

Hi Cem Dilmegani. What sets Skan.ai apart from other process mining tools is that it goes beyond mining and discovers the process without any system integration. It is an AI-powered process discovery software that uses computer vision to continuously observe each human-machine interaction and uncover every process variant or nuance. It creates a digital twin of the enterprise, also called the invisible enterprise. This twin displays all the hidden business processes that could easily be missed by human observers or by mining the enterprise systems. One can aim for automation, re-engineering, process design, precision training, and much more with this process twin created with our process discovery tool.

Kailash Chandra Jena
Nov 10, 2020 at 20:49

Thanks for the detailed description

Cem Dilmegani
Nov 14, 2020 at 15:39

Thanks!

Bhavana Agarwal
Oct 18, 2020 at 07:01

Thanks u

NewYork York Consultants
Sep 11, 2020 at 06:57

You made some good points there. I looked on the internet for the subject matter and found most individuals will go along with with your blog.

Robotic Process Automation (RPA) service providers New York

Cem Dilmegani
Sep 13, 2020 at 16:53

Thanks!

Naresh kumar m
Aug 18, 2020 at 13:45

Awesome

Sakthivel
Jun 02, 2020 at 04:42

Very useful things. Thankyou

AIMultiple
Jun 02, 2020 at 05:48

Thank you!

Rohit Bansal
Aug 20, 2019 at 10:55

Worldwide use of robotic process automation has led to significant, positive impacts on business productivity. In 2018 the adoption of robotic process automation grew globally at a higher pace than ever before. Extrapolating RPA

Ramakrishna
Jul 31, 2019 at 11:00

Thank you for sharing very useful information

BoTree Technology
Jun 20, 2019 at 12:42

Thanks for highlighting the problems that occur in most of the RPA systems, and the compiled checksheet is an add on bonus for us. Keep up the good work

Infrrd
Apr 17, 2019 at 06:06

Infrrd AI(https://infrrd.ai/) takes out the pain of capturing data and extracting intelligence from a plethora of documents, delivering you the information you need without having to manually go over documents one by one. The software basically saves you precious time and effort and you never have to worry about accuracy as Infrrd AI uses the latest OCR technologies and their own AI algorithms to ensure the precision of every extracted data.

bdevils464
Feb 20, 2019 at 11:54

Thanks for the Uipth Training information provided to me.

Oliver Harris
Aug 09, 2018 at 09:21

In order to implement RPA

Related research