See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/369201717

Prototyping and Implementing Robotic Process Automation in Accounting

Firms: Benefits, Challenges and Opportunities to Audit Automation

Article in International Journal of Accounting Information Systems · September 2023

DOI: 10.1016/j.accinf.2023.100641

CITATIONS READS

0 182

3 authors, including:

Eric Lee Mui Kim Chu

University of Northern Iowa Singapore Institute of Technology (SIT)
14 PUBLICATIONS 121 CITATIONS 5 PUBLICATIONS 22 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Mui Kim Chu on 17 October 2023.

The user has requested enhancement of the downloaded file.

 International Journal of Accounting Information Systems 51 (2023) 100641

Contents lists available at ScienceDirect

International Journal of Accounting

Information Systems
journal homepage: www.elsevier.com/locate/accinf

Prototyping and implementing Robotic Process Automation in

accounting firms: Benefits, challenges and opportunities to
audit automation
Arif Perdana a, *, W. Eric Lee b, Chu Mui Kim c
a
Monash University, Indonesia
b
University of Northern Iowa, US
c
Singapore Institute of Technology, Singapore

A R T I C L E I N F O A B S T R A C T

Keywords: Despite the advancements in technology, auditors still spend a significant amount of time per­
Robotic process automation forming repetitive and rule-based tasks. Our paper examines various audit scenarios within four
RPA accounting firms and discusses the potential for robotic process automation (RPA) to improve the
Accounting firms
efficiency and accuracy of these tasks. We propose and implement RPA-enabled solutions to make
Audit process
Audit automation
various procedures more efficient and effective. These anecdotes were based on our consultancy
Prototype experiences with auditors from a group of Big 4 and mid-sized accounting firms. We present four
Implementation practical business process scenarios and the prototyped RPA solutions. In each scenario, we
explain our automation strategies and the duration of development. We also describe the
implementation benefits and challenges, as well as analyze how we have helped the accounting
firms automate their real-world audit tasks by providing the as-is and to-be Business Process
Model and Notation that one can extend for future RPA development. Finally, we discuss the
future challenges and opportunities regarding the broader implementation of RPA in accounting
and audit tasks.

1. Introduction

Robotic Process Automation (RPA) is a form of automation technology that uses software robots to automate business tasks
typically performed by humans. RPA uses logic-driven software applications and structured-input programs to execute designated
digital labor tasks at the user interface level (Plattfaut and Borghoff, 2022). Contrary to popular beliefs, RPA does not involve physical
robots performing operational processes. Instead, digital bots that can be easily configured and trained to the various tasks are used
(Kokina & Blanchette, 2019; Moffitt et al., 2018; Zhang, 2019). As these bots can operate 24/7, it essentially eliminates any downtime-
induced wastage (Huang and Vasarhelyi, 2019).
With RPA, one can deploy software bots to capture and manipulate transactions, thereby triggering responses in unison with other
digital business systems. Together, they help to process and settle these transactions (Syed et al., 2020). Besides helping to reduce labor
costs, RPA also enables businesses to reduce human error. Overall, the RPA system is a low-cost, ready-to-implement technology that
requires minimal information technology (IT) infrastructure integration to operate (Huang and Vasarhelyi, 2019). This is important in

* Corresponding author.
E-mail addresses: [email protected] (A. Perdana), [email protected] (W.E. Lee), [email protected] (C. Mui Kim).

https://doi.org/10.1016/j.accinf.2023.100641
Received 8 April 2022; Received in revised form 9 March 2023; Accepted 20 July 2023
Available online 31 July 2023
1467-0895/© 2023 Elsevier Inc. All rights reserved.
A. Perdana et al. International Journal of Accounting Information Systems 51 (2023) 100641

today’s workplace as companies strive to streamline both business operation and employees’ mobilization (Schatsky and Muraskin,
2016). However, RPA may not be suitable for all types of business enterprise tasks (Moffitt et al., 2018). In the case of more complex
tasks, implementing intelligent process automation (IPA) – a combination of RPA and artificial intelligence (AI) – may be more
suitable. Using multiple machines to complete the tasks and keeping human intervention in the loop can more effectively optimize the
automation process (Zhang, 2019).
In the accounting domain, RPA has been gaining traction in both academia and recent practice (Bakarich and O’Brien, 2021;
Cooper et al., 2019; Kokina and Blanchette, 2019; Moffitt et al., 2018; Zhang, 2019). A few accounting firms have started to adopt RPA,
automating their mundane and repetitive business tasks in such areas as bank reconciliations and analytical procedures (Moffitt et al.,
2018). These accounting firms have witnessed first-hand the efficiency gains from implementing RPA (Cooper et al., 2019), and have
further plans to leverage this momentum into other business avenues.
Unlike the 4-stage framework (i.e., procedure selection, modification, implementation, as well as evaluation and operation)
proposed by Huang & Vasarhelyi (2019) in regard to the application of RPA in audit practice, our study provides actual business
process scenarios from a broad spectrum, describing the problems facing four Big-4 and mid-tier accounting firms located in an Asia
Pacific country, and the prototyped RPA solutions in resolving these problems. All of these anecdotes were based on our recent
consultancy experiences with auditors from these firms. In each scenario, we outlined the real-world audit tasks that the different
accounting firms were trying to automate, followed by the various automation strategies that we have designed and subsequently
implemented to help them resolve those issues.
Our study responds to Plattfaut & Borghoff’s (2022) call to evaluate those accounting and audit processes that can benefit from RPA
adoption. Prior work has focused on the developmental concept, evaluation framework, and the adoption process (e.g., Eulerich et al.,
2022; Huang and Vasarhelyi, 2019; Kokina and Blanchette, 2019; Moffitt et al., 2018; Zhang et al., 2022). In particular, the framework
developed by Eulerich et al. (2022) helps guide auditors in deciding the type of activities that would benefit from RPA. Zhang et al.
(2022) further used a case-study approach to trace the beginning-to-end processes of actual RPA implementation in the contexts of an
industrial firm and an accounting firm. However, research documenting the actual RPA prototype-to-implementation use cases is
rarely published in academic journals.
Our work complements previously published RPA studies in the accounting domain by showcasing actual collaborations among
faculty, students, and the audit firm personnel in devising viable RPA solutions for a variety of audit tasks. We present four practical
business process scenarios and the prototyped RPA solutions. For each scenario, we provide the as-is and to-be Business Process Model
and Notation that one can extend for future RPA development. We also describe the implementation benefits and challenges, as well as
explain our automation strategies and the duration of development. Complementing previous theoretical contributions, our paper
primarily aims to provide accounting scholars and practitioners with a better applied understanding of how a wide range of audit
processes can benefit from real-world RPA implementations as well as some of the potential pitfalls to avoid. The implementation
contexts as prototyped by the different scenarios are aimed at ultimately benefiting accounting firms as the latter consider broader RPA
incorporation in a wider range of tasks.
From the perspective of IT governance framework (Wilkin and Chenhall, 2020), we contend that RPA implementation should be
supported by an effective mechanism of governance. Organizing and implementing RPA governance methodically ensures that IT will
align with business processes. Selecting and deploying appropriate tools and processes enhance the performance and security of in­
dividual RPA projects, which in turn also extend to that of the overarching enterprise-wide automation landscape. The goal is thus to
efficiently orchestrate the collaboration of processes, people, and bots across an enterprise so that they can all be in line with the
automation strategy. In addition, the organization should integrate its data governance program with its IT governance initiatives, so
that the RPA data can be both readily and securely managed.
This paper is organized as follows. In the following section, we explain the current landscape of RPA in accounting. From there, we
describe the application scenarios in the contexts of the different accounting firms. For each scenario, we propose a prototype and
deploy the RPA implementation plan to help automate the business task within the accounting firm involved. We then discuss the
implementation benefits and challenges, as well as the duration of development for each RPA scenario. Finally, we provide our an­
alyses of future RPA challenges and opportunities, and draw our implication and conclusion.

2. Current landscape of RPA in accounting

In accounting, the field of auditing stands out as one area that is ripe for RPA implementation. Audit procedures consist of a set
sequence of processes that is used to determine the type and quality of information going into the financial statements (Leung et al.,
2015; Thibodeau et al., 2017). Given that the audit process entails handling many repetitive tasks, RPA can leverage its speed and
agility to help automate this time-consuming work, thereby redesigning processes to ultimately increase efficiency and effectiveness
(Huang & Vasarhelyi, 2019; Moffitt et al., 2018; Schatsky and Muraskin, 2016). Some of the audit tasks that can be automated with
RPA include: (1) Perform reconciliation of cash transactions, sales and purchases, journal entries, inventories, payrolls, and other
expenditures; (2) Monitor risk thresholds and generating alerts when they are exceeded; and (3) Other processes that require minimal
human judgment but are time consuming and repetitive (Cooper et al., 2019; Huang and Vasarhelyi, 2019; Moffitt et al., 2018;
Schatsky and Muraskin, 2016).
The process of obtaining sufficient and appropriate audit evidence requires the auditor to perform one or a combination of audit
procedures such as observation, inspection, confirmation, recalculation, reperformance and analytical procedures (Thibodeau et al.,
2017). Inspection of documents and records include such processes as vouching and tracing. Vouching involves the examination of
transactions recorded in the books of accounts by reviewing the vouchers received (for example, invoices, receiving reports and

2
A. Perdana et al. International Journal of Accounting Information Systems 51 (2023) 100641

purchase orders), so as to tally with the entries made in the accounting records, verifying whether there are any errors in the amounts
recorded and also ensuring that transactions are recorded in the correct accounts (Leung et al., 2015). This document review process is
often repetitive, labor intensive and time-consuming, thus presenting an ideal set of conditions for one to consider deploying RPA
(Huang and Vasarhelyi, 2019).
By implementing RPA to improvise routine activities, accounting firms can effectively increase capacity at all levels. Using RPA also
allows auditors to perform tests on the entire population of accounting data, instead of the usual method of manual sampling, thus
effectively reducing the number of potential sampling errors (Huang and Vasarhelyi, 2019). The use of RPA, when combined with
analytics, can lead to better detection of accounting anomalies, discrepancies and even fraud (Zhang, 2019). One can verify that the
data entered in the accounting records is complete and free of duplicates by performing an automated comparison with the source
documents. Further efficiency can be achieved by continually monitoring and notifying users when the risk thresholds are exceeded
(Moffitt et al., 2018).
Every successful audit should begin with a well-conceived planning phase, one in which the auditor gathers information about the
client, including its key business tasks. This will ensure a deeper understanding of the client, thereby enabling the auditor to decide on
the extent of risk assessment and planning processes to be performed (Leung et al., 2015; Thibodeau et al., 2017). Automation is not
typically implemented at this stage because the process of determining the key risk areas, materiality, scope, and strategy for an audit
requires professional judgement and is not one that can be easily automated. One will generally incorporate automation only after the
planning phase, when it can be used to support the fieldwork, testing controls and details of balances, disclosures, as well as the
underlying transactions associated with the client’s financial statements (Thibodeau et al., 2017). Testing of both controls and details
possess the inherent characteristics that make them suitable for automation. Both processes entail repetitive activities that involve a
high level of documentation (Huang and Vasarhelyi, 2019).
Both tests of controls and details involve the basic process of inspecting the supporting documents and records. RPA tools enable
auditors to automate the process of gathering information by capturing audit-relevant data from source documents before presenting it
in a standardized format within the audit working papers. Audit-relevant data includes such information as customer name, address,
invoice number, invoice date, and invoice amount. Data analysis software is used to import the recorded data from accounting records,
which can then be matched with the audit-relevant data collected with RPA using pre-programmed audit tests. RPA, therefore, helps
facilitate auditors’ work with more streamlined access to information (Kokina and Blanchette, 2019).
RPA, when left unattended, may not be amenable to all audit tasks (Zhang et al., 2021). Some audit tasks require judgement and
auditor interventions (e.g., validation and verification). Attended RPA that requires a certain level of auditors’ judgments and vali­
dation of RPA outputs can present more use cases in the audit domain. By capturing a more detailed level of data, RPA allows the
auditors to more accurately assess the risk of material misstatement in the financial statements. The auditors can hence more judi­
ciously exercise their judgments regarding the generation of RPA reports. Inclusion of more reliable information enhances business
value, which in turn provides stakeholders with a higher level of assurance (Huang and Vasarhelyi, 2019).

3. Description of scenarios and prototyped RPA solutions

At the initial phase of our consultancy, the RPA solutions were first conceived as proposed plans through a series of collaborative
brainstorming exercises amongst the firms’ personnel, the students and us, that focused on key (and potentially repetitive) audit task
areas. Upon completing detailed discussions of the plans with the senior audit teams from the firms involved, the RPA prototypes for
the various use cases were eventually implemented within the accounting firms. In the course of implementation, our students (as our
liaison and as part of their work engagement stint with these firms), together with the audit personnel involved, had a firsthand view
and hence were able to help monitor the implementation through periodic progress reports. Throughout the entire implementation
process, the students, firms’ personnel and us also regularly convened, so as to promptly resolve any outstanding issues as they arose.
Generally, within their first few months at the firms, the students worked together with us to identify/re-affirm those various key
areas that were found to entail repetitive tasks, and from there we developed suitable RPA use cases for automation in those areas. In
consultation with the audit personnel involved, all of the three parties then held numerous meetings to discuss how various RPA
solutions could be implemented in different scenarios, so as to improve audit effectiveness and/or efficiency. Upon subsequently
getting approval from the firms’ senior teams, we promptly deployed the prototyped RPA solutions in the different functional pro­
cedures/areas.
We implemented all of our bots in a decentralized fashion, which means that the bot governance was all at an individual or team
level. Here, the individual business process areas owned and managed all aspects of the process, including governance, opportunity
assessments, build, test and deployment, as well as operations, with ad hoc coordination among the process owners (see for e.g.,
Deloitte, 2019). Consequently, these bots’ deployment structures were largely in silos. The accounting firms adopted the RPA solution
by deploying the bots during our consultancy period. During the course of implementation, for around a five-month period, we
communicated regularly with the firms’ personnel. On average, each bot was used for at least five engagements during the consultancy
period. We were, however, unable to track which engagement(s) and how the RPA solution was used after this period.
Overall, these initiatives emerged from our consultancy work, with the aim of showcasing the RPA solutions in adding substantive
practical value and make the work assignments of our students more efficient/effective. Importantly, we believe that such application
offers an applied research relevance related to the complexity of bot deployment in accounting firms by leveraging how students, as the
primary stakeholders in academia, connects with our industrial partners. From the firms’ perspective, they could evidently witness the
implementation outcomes firsthand, hence achieving the practical synergy of staying engaged with academe toward making further
RPA inroads in other audit procedures. In the following subsections, we provide details of the four scenarios, including explanations of

3
A. Perdana et al. International Journal of Accounting Information Systems 51 (2023) 100641

each scenario’s as-is and to-be business processes, as well as the implementation benefits and challenges.

3.1. Scenario 1: Employee loan Auditing: Vouching, adherence and recalculation

A typical business process in an accounting firm ranges from audit planning to the reporting phase, and finally to the completion of
the audit. The professionals involved in this process are mainly audit partners, managers, seniors, and associates. The most labor-
intensive and time-consuming part of auditing is the performance phase. This phase is the main part of the external audit, and
typically involves tasks like testing a client’s ability to correctly capture and process data in the preparation of its financial statements.
Without automation, an accounting firm will normally perform this process by reviewing and manually extracting data from docu­
ments created by its audit clients, so as to ascertain that the data were captured correctly. At this stage, additional files and records may
be requested by the auditors.
In this scenario, auditors at Accounting Firm A (AFA) request documents and audit schedules through a dedicated portal, by
providing encrypted login credentials to the clients that enable them to upload the requested documents online. This portal permits the
auditors to communicate directly with their clients when they request supporting documents, and it helps to keep track of the number
of requests outstanding to date. In addition, the portal serves the function of sending automatic email reminders, so that a client is
notified when a request is made as well as when a request deadline is approaching.
After the client uploads the requested documents, the auditors have to perform the time-consuming tasks of analyzing the volu­
minous data, selecting relevant samples, and sifting through the relevant data source. Even if the auditors use spreadsheet as part of the
audit, these massive and repetitive tasks are still time-consuming and inefficient. A potential case in point here is employee loan testing
(ELT) at AFA. The goal of ELT is to verify that amounts due have been correctly identified, measured, recorded and reported in the
financial statements. This is an audit area that could potentially benefit from implementing RPA due to the repetitive tasks involved.
The ELT process includes various activities that do not require audit judgement but are nonetheless time-consuming. Such activities
include collecting and preparing audit evidence, as well as performing mechanical tasks like reconciling loan balances and verifying
the interest rates from one data source to another. Without automation, auditors must manually recalculate, track down relevant
documents, and verify compliance with the terms and conditions of loan contracts. As shown in Fig. 1 in the Appendix, since there are
multiple tests of details to be performed and many documents and data sources to be rechecked, this process often results in many
errors and inefficiencies.
Using the sampling methodology, the vouching process at AFA currently consists of requesting the necessary documents from
clients and manually filling out the sample size form. This includes specifying such attributes as account name, coverage period,
corresponding performance materiality and population size. The respective tolerable misstatements will then be defined, so as to
generate the minimum required samples. This will in turn allow the auditor to manually and randomly select samples from the master
file provided by the client. After selecting the specific samples, the auditor will then communicate with the client and request sup­
porting documentation to ascertain and verify the details. Upon completion, the auditor will then enter the required demographic
information for each sample into the audit working papers to serve as evidence and to facilitate subsequent referencing.
The main problem with this process is the large amount of work involved, often resulting in an enormous amount of time spent on
vouching that could instead be spent in other audit areas. Depending on the nature and complexity of the client’s operations, the audit
often requires a large sample, which can increase when the materiality threshold drops, or when the account has a higher risk of
misstatement due to weak internal controls. Overall, it may take a substantial amount of time for an auditor to scan the document and
enter the required information into the working papers, followed by determining if the samples contain any misrepresentations. This
valuable time could instead be spent on other essential tasks, such as analyzing the results of audit tests for material misstatements. In
addition, with the large volume of samples needing to be analyzed, human errors can result, leading to erroneous information being
entered into the working papers.

3.2. Prototyped RPA solutions for scenario 1

When RPA is implemented in this scenario, auditors need only request the documents and statements through the dedicated online
platform that connects the firm and the client. This ensures that the data received has standardized labels and is consistently formatted,
so that RPA bots can perform the vouching, recalculation, and compliance testing. Once the client-specific data is entered, auditors
need only to standardize the form and make them machine-readable so that RPA bots can read the data. Based on the uploaded
documents, RPA bots will then create an audit working paper in the dedicated online platform so as to streamline the testing of such
work steps as vouching, recalculation and compliance testing.
The entire process can be expedited as it will be transformed into a more effective and efficient workflow for substantive testing.
Like the current manual testing, the auditor will first request the appropriate master files with relevant reports from the clients. Unlike
the traditional method, whereby the file received can be in any format, from soft copy to hard copy, the files required for automating
the vouching process must be digitized and in a specific format. Once the documents are received, depending on the structure of the
data that is provided by the customer, the auditor may need to perform the additional step of data cleansing to ensure that the inputs
and their relevant attributes are in a format that the program can read. For this purpose, RPA bots can be pre-programmed to allow the
auditor to validate the texts and numbers before this information are subjected to further processing. From here, the auditor can run
the configured program instead of manually selecting samples. The samples are automatically selected and vouched by identifying the
most salient terms previously defined, so as to match them with the corresponding source documents.
After selecting the samples, RPA bots can automatically request the documents required for vouching from the client by sending

4
A. Perdana et al. International Journal of Accounting Information Systems 51 (2023) 100641

Table 1
Estimated Time on RPA Development for Scenario 1.
Initial understanding of the business process Proposed automation for the business process Bot development Bot testing

1 month 0.5 month 1 month 0.5 month

emails or notifications through the online platform. Once the client uploads the documents to the appropriate online platform, RPA
bots perform the programmed tasks and complete the audit working papers. For example, any disbursements made need to be verified
with a disbursement register and an account statement. RPA bots can thus be pre-programmed to match the records from the
disbursement register to the bank and account statements, thereby ensuring that proper disbursements are made over the year. Any
discrepancies will be annotated for the auditor to review, and to take further action if necessary. In essence, this means that automation
has simplified the traditional way of manual vouching to a more streamlined series of steps. Consequently, the vouching process can be
expeditiously completed with a summary of the relevant information, accompanied by the appropriate supporting documents. With
automation, attributes that match the description of each sample can be quickly identified by searching the database library. In
addition, attaching the supporting documents makes it easier for auditors to look them up during the assessment, as they do not have to
spend time searching for the location of each document. Auditors can then ensure that a pre-specified checklist has been fulfilled and
thereupon form a judgment about possible discrepancies indicated by the program.
Once verification is complete, data that matches records will be marked “Yes” to indicate that the document exists and has been
adequately accounted for. Records with errors are marked “No,” with a notation in the ”Remarks“ column for the auditor to follow up.
Legal compliance refers to compliance with the legal requirements outlined in the law or in the contracts. The automation process will
enable the referencing of the Financial Reporting Standards and the client’s legal documents, so as to look for keywords that the
auditor will otherwise look for when considering compliance. Similarly, for vouching, a ”Yes“ means no further action, while a ”No“
means further attention is required by the auditor.
Evidently, the current system of vouching, whilst cumbersome and time-consuming, is indispensable since it forms the ’backbone’
of an audit. Integrating technology is the optimal way to circumvent this tedious task while ensuring that it is done accurately and
efficiently. By automating the vouching process, the amount of auditor intervention can be kept to a minimum. The primary purpose is
ultimately to monitor and to regularly maintain the program. Additionally, RPA bots can be programmed to write formulas in
spreadsheet and thereupon be transferred to the AFA’s audit working paper for recalculation. For example, auditors can ensure that
none of the loans is impaired or that the recoverable loan amount is not less than its carrying amount. Based on the client’s docu­
mentation, some prior adjustments may be required in order to facilitate the recalculation. Last but not least, RPA bots can help input
necessary data into the working papers, such as when recoverable amount is less than the carrying amount. In addition, the recal­
culation of interest income should be adjusted depending on the interest rate specified in the loan agreement and the number of days
for which interest needs to be calculated. Fig. 2 in Appendix details the automated business process when RPA is implemented for this
scenario.

3.3. Implementation benefits and challenges for scenario 1

The ELT in AFA, without RPA, typically needs a turnaround of 1 to 1.5 days to process about 20 items by an auditor. This estimate
does not include the time required to request relevant documents from the clients. If RPA is implemented, four hours can be saved. The
time savings are mainly due to automating the execution of impairment tests and recalculating interest income (see Fig. 2 in the
Appendix). These two tasks can be automated by enabling RPA bots’ access to spreadsheets. While the automation prototype can save
time, the items to be evaluated during impairment testing and the recalculation of interest income must be manually entered into the
spreadsheet form after the loan has been inspected and reviewed by the auditors. This gave the auditors the impression that the bot
cannot fully automate the tasks as yet. The challenge in developing this automation solution for ELT primarily revolves setting up the
bots to properly access the ELT related documents, spreadsheets, perform the calculations using the stipulated spreadsheet formulas, as
well as input the results of the valuation and calculations into the auditors’ working papers. This development process requires careful
application integration (for e.g., in the areas of processing range and tables within spreadsheets) and experimentation with the existing
documents to ensure that the calculation is accurate, and that the output is in the correct format. During implementation, there were
some technical glitches with the application integration due to the different screen resolutions from multiple laptops running the bots
simultaneously. A few auditors also felt that some of the bots can be inherently problematic and hence require prior adjustments.
Overall, the time to develop the automation solution for ELT is estimated to take about three months, as shown in Table 1.

3.4. Scenario 2: Bank reconciliation Checks, analytical procedures and Three-Way match

There are several business tasks that involve auditing within Accounting Firm B (AFB). Three of these tasks are amenable to RPA
automation: bank reconciliation checks, analytical checks and three-way reconciliations.

3.5. Scenario 2A: Bank reconciliation

The main purpose of bank reconciliations is to compare the cash account balances in an entity’s general ledger (GL) with the

5
A. Perdana et al. International Journal of Accounting Information Systems 51 (2023) 100641

Table 2
Estimated Time on RPA Development for Scenario 2A.
Initial understanding of the business process Proposed automation for the business process Bot development Bot testing

0.5 month 0.5 month 0.5 month 0.5 month

corresponding figure indicated on the bank statements within the same time period so as to check for consistency, accuracy and
completeness. Bank confirmation is an essential procedure that complements the auditing of bank reconciliations. As the name sug­
gests, this task seeks to obtain confirmations from the banks on the year-end cash balances and on the clients’ loans outstanding. Fig. 3
in the Appendix outlines the procedures followed by an auditor in AFB, who is tasked to complete the bank reconciliation check and
henceforth request bank confirmations for the financial year-end audit engagement. The auditor begins the process by requesting the
client to provide the signed bank confirmation request letters that will in turn be sent to the banks, along with the year-end bank
statements, the GL and clients’ bank reconciliation statements.
After the client has provided the requested documents and records, the auditor sends the bank confirmation requests (note: this
request can take the bank a few days to reply) and concurrently performs a comparison between the reconciliation statements and the
corresponding bank statements, before documenting the reconciled figures in the working papers. At this point, the auditor must
ensure that the reconciled figures tally with the figures on the GL. If the auditor finds any reconciliation item dubious or needs further
clarification, he/she will enquire with the client on the nature of the reconciliation item and note it on the working papers. Upon
receiving the bank confirmation reply from the banks, the auditor then compares the figures indicated on the bank confirmations with
the reconciled figures documented earlier. If there are no issues, the auditor completes the process by documenting with check marks
and/or some short comments, and then uploads the working papers into the audit system. However, if the auditor discovers any
disparity between the two figures, he/she will then be required to seek further clarification from the client either via email or a face-to-
face meeting. Ultimately, the auditor will only complete this process after he has obtained a well substantiated and satisfactory
explanation regarding any differences.
One task that can potentially be automated is the comparison of reconciliation with the bank statements, and the matching of
figures in the bank confirmation letters with the working papers. Reason for automation in both cases is because bank reconciliation is
a repetitive task that only requires the auditor to check for any difference in figures from the documents. The auditor can instead
expend his/her time to more fruitfully complete more advanced tasks requiring judgement.

3.6. Prototyped RPA solutions for scenario 2A – Bank reconciliation

With RPA, once the file targets are defined, the bots will automatically capture the information on the reconciliation statements,
bank statements and bank confirmation letters. This can be accomplished by converting information on the files to texts through RPA
programming. Through automation, this process captures the defined information before it is transferred to the working papers. The
bots can then compare the corresponding numbers in the working papers and thus identify either large reconciling items (note: this
amount can be pre-defined in the RPA structure) or discrepancies between the numbers (see, Fig. 4 in the Appendix). The numbers are
then highlighted in the working papers for subsequent reviews by the auditors. Once again, automating manual audits can reduce the
number of repetitive tasks performed by the auditors, thereby channeling the auditors’ skills and experience toward tasks requiring
unique individual judgements.

3.7. Implementation benefits and challenges for scenario 2A – Bank reconciliation

In a regular audit season, an auditor in the accounting firm can typically analyze five to ten bank statements daily. This time
estimate excludes the amount of time required to interview clients and obtain bank confirmations. Automating bank reconciliation can
save three to four hours, doubling the number of bank statements analyzed daily. As a result, the auditor’s role can shift from manual
review and calculation, to reviewing of working papers. Developing RPA for bank reconciliation is less cumbersome than RPA for ELT.
The logical functions to check the discrepancy is written in RPA bots. These bots then execute the functions to the given inputs. The
challenge one can face here is that when RPA creates the working papers for matching, the extracted number is sometimes incorrect.
Using optical character recognition (OCR) can result in reading errors, particularly for dollar amounts with decimal places. If the
written character is unique, OCR can capture the data correctly. However, the numeric text is sometimes distorted if the character is
the result of converting either an image or a scanned document. An amount worth $1,875.25 could be erroneously read as $1,876.26.
Auditors must carefully consider these potential errors and to review the RPA-generated working papers before the figure is processed
further. As observed, the occasional inaccuracies of the OCR when reading the images or the scanned documents also left a somewhat
unfavorable impression on the auditors. Overall, the time required to develop the bank reconciliation automation solution is estimated
to be approximately two months, as presented in Table 2.

3.8. Scenario 2B: Analytical procedures

Analytical procedures are important in auditing because it evaluates both the accuracy and reasonableness objectives by estimating
the figures for predictable accounts (such as prediction of revenue accounts) and comparing them with the actual figures as reported by

6
A. Perdana et al. International Journal of Accounting Information Systems 51 (2023) 100641

Table 3
Estimated Time on RPA Development for Scenario 2B.
Initial understanding of the business process Proposed automation for the business process Bot development Bot testing

0.5 month 0.5 month 0.5 month 0.5 month

the client. This is primarily done through calculation-modeling within the audit working papers. As illustrated in Fig. 5 of the Ap­
pendix, the process starts with the auditor requesting access to the relevant information (for e.g., average number of customers per
month, prices per unit, or average number of units sold for each product category) so as to get an estimate of predictable accounts (i.e.,
revenue in this case).
In addition, the auditor will submit a request for the GL from the client for comparison purposes. When all the required information
and GL are generated, the auditor will then make use of the information to gauge the figures that are expected to be reported by the
client. The auditor thereafter calculates the performance materiality (PM) by utilizing the thresholds limit that is set by the audit firm.
The estimated figures, materiality and figures from the GL are then updated within the audit system, as well as manually transferred to
the corresponding working papers for the different accounts. If any differences between the estimated and reported figures do not
exceed the materiality level, the auditor can finalize the analytical check process by uploading the completed working paper into the
audit system. However, if an account is discovered to have a material discrepancy, the auditor will have to enquire with the client and
investigate possible reasons for any material differences. Ultimately, the auditor can only complete the task if he/she is sufficiently
convinced by corroborated evidence, showing that the discrepancy was due to legitimate reasons.
The processes that can potentially be automated are the calculations of the predicted figures and materiality, along with a com­
parison of the estimated and actual figures that are found to flag those accounts over the materiality threshold. The main reason why
these processes should be automated is because of the perceived improvement in calculation accuracy with RPA, given that humans are
more prone to computation errors. Overall, using RPA for analytical checks should be more efficient as it reduces both the possibility of
arithmetic mistakes and also the time taken to derive the figures.

3.9. Prototyped RPA solutions for scenario 2B – Analytical procedures

Without automation, the auditor must calculate materiality and estimate the predicted figures of the accounts. At the same time,
they have to manually determine whether differences between the predicted and actual figures were above the materiality threshold.
With automation, the bots can be pre-programmed to calculate materiality and to determine the estimates based on clients’ input,
whereupon the numbers can be transferred to the working papers.
With a preset RPA structure, the bots can compare the estimated and actual numbers on the working papers. Any significant
variances that exceed the materiality threshold can then be highlighted. While the auditor can be notified to review and further
investigate any differences in those areas where the revenue account is generally predictable, the bots can make further predictions of
the expected revenue. In cases where revenues can be reasonably estimated, some examples of revenue reasonableness tests include:

a. For a hotel, expected room revenue = hotel room days * occupancy rate* average daily rate
b. For a shipping company, expected vessel charter revenue = no. of days chartered * daily rate
c. For a commission-based company, expected commission income = sales receipts from point- of-sales system * commission rates

This automation process potentially reduces human errors in calculation and the need for manual, repetitive tasks. This will in turn
help to increase productivity by devoting billable time more fruitfully toward other essential tasks.

3.10. Implementation benefits and challenges for scenario 2B – Analytical procedures

The analytical procedures for the revenue account in AFB can regularly be performed by an auditor within 1.5 to 2 days. With the
implementation of RPA in the analytical procedures, the analysis of the revenue accounts can be done faster. The estimated time
savings is worth about half a day. This is achieved by having some tasks handled by bots. For example, four tasks (calculating per­
formance materiality, calculating estimation results, creating working papers, and comparing estimation results with GL) are now
undertaken by bots (see Fig. 6 in the Appendix). Before the data is processed for these four tasks, the data has to be prepared, cleaned,
and manually entered into the spreadsheet by the auditor. From our experience developing RPA in this area, challenges arose in writing
the syntax for the spreadsheet and RPA bots. The four tasks above required extensive application integration bots, as well as multiple
primary and nested decision flows. Even after the documents were received, manual tasks were still involved in this process. After the
data was processed via RPA, the auditors then reviewed the comparison documents and the final working papers. Nonetheless, with the
analytical procedure functions taken over by the bots as a result of this automated process, the auditors were able to perform less
manual work, particularly those involving repetitive calculations. Table 3 presents the time spent for the RPA development.

3.11. Scenario 2C: Three-Way match

Three-way match is used when the auditor uses three different documents to ensure the accuracy, occurrence, and completeness of

7
A. Perdana et al. International Journal of Accounting Information Systems 51 (2023) 100641

Table 4
Estimated Time on RPA Development for Scenario 2C.
Initial understanding of the business process Proposed automation for the business process Bot development Bot testing

1 month 0.5 month 1 month 0.5 month

the transactions. In AFB, three-way matches are usually used for sales and purchase accounts. Figs. 7 and 9 in the Appendix illustrate
the three-way matching processes (without RPA) for both the sales and purchase accounts respectively. While the processes for both
accounts are similar, the documents involved differ. Both processes begin by having the auditor request the client’s GL and relevant
spreadsheet files such as the purchase spreadsheet. Upon receiving the GL and spreadsheet file from the client, the auditor must check
the dataset to ensure that there are no duplicates in the file and, from there, upload the inspected file into the audit system. The system
will then select the samples for the auditor to perform vouching. At the same time, the working papers need to document the selected
samples and the relevant results. Upon completion, the auditor will request for the purchase orders, invoices and order receipts related
to the samples selected. For the sales transactions, the auditor will need to separately request from the client the sales order, invoices,
and the delivery orders.
Once the auditor receives the full set of documents from the client, he/she will proceed to do the three-way matches and fill in
details of the documents within the working papers. If there is no problem, the process will end after the auditor documents the results
in the working papers within the audit system. On the other hand, if any issues are noted, the auditor is required to conduct a client
inquiry and seek corroborating explanation that no fraud has been committed.
The processes that can be automated under the three-way matching procedure are the sending of request emails to the clients and
the actual matching of the relevant documents requested. Request emails that are sent by the auditor for the purpose of three-way
matching are often repetitive and have similar contents. It is thus possible to integrate the relaying of sample information with an
email automation process. Another benefit of utilizing RPA here is to minimize human errors arising from matching the different
documents. This is a problem that is particularly prevalent during the busy seasons, when fatigue commonly sets in as auditors face
tight deadlines. Thus, to save time and to minimize occurrence of human errors, it is advisable for one to automate the matching
process.

3.12. Prototyped RPA solutions for scenario 2C – Three-Way match

Currently, the auditor needs to start the triple-matching process by sending emails to the client in regard to the samples and the
matching of documents related to those samples. With the automation of this process, bots will instead replace the auditor in sending
the request emails involving similar contents, for the purpose of triple-matching. All the auditor needs to do is to preset the email
recipients in the RPA structure. Once the three different documents are received, the bots can then proceed to automatically capture
the information from the pre-defined target files. This automation uses programming to convert the information into text before
subsequent transfer to the working papers for automatic matching. When any difference is highlighted, the bots will prompt the
auditor to examine those areas of concern. Such RPA implementation will end up saving the auditor hours of matching work that are
typically done manually (see Figs. 8 and 10 in the Appendix for the business processes after automation).

3.13. Implementation benefits and challenges for scenario 2C – Three-Way match

An auditor in AFB can typically analyze 40 items per account within four hours. With the help of RPA bots, the estimated time
savings is two hours. With the automation solution, the three-way match can be performed faster for either sales or purchasing
transactions. Within four hours, we saw that the number of items analyzed was doubled. The challenge with this automation was to
extract the data from different forms of documents (some of them were digitized, others were paper invoices). We had to scan the paper
invoices and convert them into a PDF format that could be further processed with OCR. The extracted data was then automatically
entered into the spreadsheet form for further processing. As in the case of ELT, OCR can sometimes produce inaccurate results.
Therefore, auditors must check the results of the extracted elements in the working paper so as to ensure their accuracies.
In retrospect, when using OCR to read images and the scanned documents, the auditors reacted unfavorably to those instances
when inaccuracies were encountered. The development of the bots for this automation includes application integration with the
spreadsheet, accessible formulas written in the spreadsheet, and primary and nested decision flows syntaxes in the RPA, so as to
evaluate the three documents within the business processes of purchasing (purchase orders, goods receipt bills and invoices) and sales
(sales order, delivery orders and invoices). Overall, the estimated time required to develop this automation solution for triple matching
was nearly three months, as presented in Table 4.

3.14. Scenario 3: Search for Unrecorded liabilities

There are several audit procedures that Accounting Firm C (AFC) can consider automating. One of the current audit procedures
performed by AFC is the search for unrecorded liabilities (SFUL). This activity is typically performed by junior staff members of the
engagement team, such as interns and associates. It is highly repetitive, which when combined with the sheer volume involved, results
in a very mundane and laborious activity. Purpose of a SFUL test is to verify, based on the date of services rendered or the goods

8
A. Perdana et al. International Journal of Accounting Information Systems 51 (2023) 100641

Table 5
Estimated Time on RPA Development for Scenario 3.
Initial understanding of the business process Proposed automation for the business process Bot development Bot testing

1 month 0.5 month 1 month 0.5 month

delivered, if the liabilities are understated as well as to ensure that any material payment vouchers issued after the client’s financial
period (or any unpaid supplier invoices) are recorded in the correct period. SFUL testing generally extends to a population sample
beyond the last day of the financial period.
As shown in Fig. 11 of the Appendix, the current SFUL process performed by AFC is as follows: First, the audit team will obtain the
GL from the audit client for the financial year, as well as any subsequent transactions made after the financial year end. With the GL
showing a list of all the transactions that have occurred, the audit team has to ascertain that the recorded expenses relate to the correct
period. After obtaining the files in spreadsheet format, the auditor will use the filter function within the spreadsheet to filter items as
per the Summary of Audit Differences (SAD) threshold amount, latter normally set by the senior team member. Additionally, during
the SFUL process, the auditor will extract invoices that are paid in the subsequent period as well as any unpaid supplier invoices,
followed by ascertaining that the procedure for sampling is similar to that of filtering for the GL transactions. This process is then
repeatedly performed for various other accounts like trade payable, cost of sales and expenses.
The auditor typically conducts the filtering process manually. One can potentially implement RPA here for the purpose of
increasing the efficiency of the procedures and processes. For example, the audit firm can develop RPA that works to supercede the
step-by-step actions typically performed by the staffs, so as to ensure data can be automatically accessed from the various spreadsheet
files. This will thus improve the overall efficiency of the filtering process, while eliminating human errors and biases.
After selecting the samples, the auditor proceeds to vouch for samples of invoices and compare them with all the receipt reports,
suppliers’ invoices, and purchase orders. This helps the auditors search for unmatched documents and identify the dates of receipts.
Whether the goods and services are considered as liability will depend on whether they are accepted before or after the year end.

3.15. Prototyped RPA solutions for scenario 3

With the help of RPA, instead of having to perform manual filters for items that are above the SAD amount across all client’s firms,
the staff will only be required to perform a few actions. Also, as the output of filtered results can be expeditiously exported to a separate
spreadsheet file for review, this file can in turn be forwarded to the client in either hard or soft copy, if the need to request for any
required supporting documentation arises.
Further, the auditors at AFC can read and retrieve data from various documents, including unstructured data, by employing IPA to
enable document-extracting via the use of OCR. If this process is automated, it can improve data accuracy, thus resulting in a higher
degree of confidence. This is because a larger scope can now be performed with the help of automation. This will provide a better sense
of reassurance to the client’s stakeholders, thereby improving audit quality.
Through OCR, a variety of document types such as scanned paper documents, PDF files, or images taken with a digital camera into
editable and searchable data can be processed. This technology makes it possible to digitize documents and make them more accessible
and usable. The use of IPA for document extraction allows the auditor to configure rules and formulas. With data extraction automated
with the use of the built-in OCR devices in IPA, auditors can now be released from vouching tasks. All the extracted data can be
exported to a spreadsheet, with fields such as the document number, date of record and the amount paid or unpaid included. Any
concerns regarding the unstructured data can also be addressed with machine learning (ML), by transforming all the digital copies of
supporting documents into a standard format, per the proposed framework of the audit data standard. Additionally, constraints can be
preset to flag out any items that are mismatched. For instance, invoice numbers that cannot be found from the prior and current years’
GL can indicate that this item is a liability not yet recorded in the system. This will thus be flagged down and the output is exported to a
separate spreadsheet or text file for the auditor to review.

3.16. Implementation benefits and challenges for scenario 3

An auditor in AFC can typically spend two hours on an SFUL. With RPA, the work can be done in one hour. RPA bots can help
auditors review payment vouchers, vendor invoices, any mismatched receiving reports, and also review the descriptions of when
services and goods were received in order to determine if they were received before year-end (see Fig. 12 in the Appendix). Similar to
Scenario 2, extracting the data from the scanned documents may introduce some errors and such occurrences were perceived unfa­
vorably by the auditors. However, these inaccuracies are often flagged by auditors before the bots proceed with the valuation and
calculation. The bot’s development for this automation involves writing the syntax in the RPA and spreadsheet application. The
estimated time to develop this prototype is presented in Table 5.

3.17. Scenario 4: Audit of statutory records (a.k.a. Stat Audit)

At Accounting Firm D (AFD), stat audit is a component within an external audit that is conducted to confirm that the client’s entity
has complied with the legal requirements relating to the company’s secretarial records. These records include register of directors,

9
A. Perdana et al. International Journal of Accounting Information Systems 51 (2023) 100641

register of charges, register of members, directors’ resolutions, minutes of meeting and business registration submitted to the gov­
ernment, such as register of charges, shares, appointment and resignations of directors or secretaries. These statutory records are either
maintained by the appointed corporate secretary in the designated corporate secretary firms or are kept by their corporate secretarial
departments.
The auditor often needs to arrange prior appointments before being physically present in the corporate secretarial firms to review
the clients’ statutory records. The auditor typically obtains physical copies of client’s statutory records from the corporate secretary-in-
charge. Upon retrieving the client’s records, the auditor will make copies before performing further audit work. Using information
retrieved from the client’s statutory records, the auditor analyzes and makes references on them. Thereafter, the audited and refer­
enced records will be scanned and filed within the audit system. The auditor is also required to summarize the resolutions or minutes
into the “summary of minutes” section of the audit working papers in assessing any audit implications (see Fig. 13 in the Appendix).
Also, registered business information to the local authority such as register of charges, secretary and shares will be reviewed and
recorded into the “statutory checklist” section of the audit working paper so as to ascertain that the business information is current and
accurate. Additionally, the auditor will contact and verify with the client, so as to ensure that all records are complete and have been
furnished by the corporate secretary. One thing to note though, with the recent COVID-19 pandemic disrupting various business
operations and many businesses having to implement “work from home” policy, many auditors have experienced temporary prohi­
bition from conducting in-office stat audits. With delays in stat audit work, this has inevitably resulted in disruption to the audit
workflow.
Many corporate secretarial firms have resorted to disseminating client’s information and statutory records over such online sharing
platforms as emails and SharePoint. For instance, during an engagement, a corporate secretary can be emailing copies of client’s
statutory records via email. As such mode of transmission is typically unencrypted, this can potentially result in confidential data
leakage if incorrectly sent to any unintended recipient, thus resulting in undesirable financial impact to the client. Overall, one should
be mindful to the fact that online sharing platforms are prone to frequent system hacking and data phishing. Evidently, this can make
the corporate secretary firms susceptible to confidentiality breaches and data privacy lapses, particularly if the client’s confidential
information is maintained via manual filing.

3.18. Prototyped RPA solutions for scenario 4

With RPA implementation, the audit procedures for stat audit will proceed from traditional methods to digitalization. Unlike the
current practice of making physical visits to the client’s place or of sharing data through emails or SharePoint, automation can aid in
reducing such burden and in enhancing the quality of audits. The accounting firm will be able to introduce its own online platform,
connected using its own private cloud, whereby it can ensure the safety and privacy of the data and information. An automated
statutory audit process will be able to connect the corporate secretary’s filings into the shared platform and hence verify the validity of
the records with the client’s information. Auditors will thus be able to access the records conveniently and also instantaneously transfer
information onto the audit working papers. Through this virtual platform, it connects both the auditors and clients, and brings all of the
data together in one place. The power of cloud computing also allows the auditors to seamlessly coordinate and manage the statutory
audit process in a consistent manner. It also streamlines communications with clients, enabling them to upload documents to the cloud,
thereby allowing auditors to keep track of the progress in real time.
As there are a lot of documents for the auditors to review, RPA bots can be deployed within the cloud interface to validate the
documents uploaded. If the dates are not within the period that are requested by auditors, a notification will be sent to request for an
updated document. Notifications will also be sent to inform auditors on the progress. The involvement of RPA bots in this process is
vital as it helps to minimize time spent gathering documents and other administrative matters. Instead of the auditors manually
organizing the documents, one can pre-program the bots in the clouds to simultaneously organize the multiple files based on specific
rules (for e.g., grouping based on time period, subsidiaries or balance sheet categories) once the documents received through the
clouds are confirmed to be within the period requested. To ensure consistency, the bots can also make sure that all the data received are
in the right format and prompt for any corrective actions if required.
With all the documents in the same format, the bots in the cloud can follow up by extracting data using IPA in a way that is useful for
the auditors. Auditors can specify the rules, such as a certain page number range, or a particular date of meeting. With just a few clicks,
one can then extract and save a copy under a specified folder used by the auditors for further analyzing and auditing. This eliminates
the need for the auditors to manually scroll through the pages to find a certain piece of information. However, when implementing
RPA, it is crucial to take note of the security risk involved. It is important to consider the possible limitations of data-sharing over any
online platforms. Data security is critical in maintaining an audit firm’s reputation and upholding its ethical and legal responsibilities.
The cloud contains many personal and confidential data of the clients, and it is important that this information not be misused. To
ensure a safe and convenient data cloud sharing environment, the accounting firms should implement data encryption or data
encoding. To reduce such threat, regular supervision, and maintenance of audit logs regarding the bot activity should be performed so
as to verify that the bots are operating within any intended set of rules. While it is necessary to encrypt data, it is also crucial to impose
access restrictions to the platform, so as to prevent any unauthorized access.

3.19. Implementation benefits and challenges for scenario 4

The stat audit is usually conducted by an auditor at AFD and lasts one and a half to two days. Some stat audit tasks, such as
evaluating procedural changes, evaluating areas of compliance and noncompliance, evaluating internal audit reports, and evaluating

10
A. Perdana et al. International Journal of Accounting Information Systems 51 (2023) 100641

Table 6
Estimated Time on RPA Development for Scenario 3.
Initial understanding of the business process Proposed automation for the business process Bot development Bot testing

0.5 month 0.5 month 0.5 month 0.5 month

Table 7
Practical Findings from the RPA Development.
Practical Implications Descriptions

Communications and the automation champion RPA developers need to be in constant communication with all affected departments. Personnel in the
latter need to provide periodic performance feedback to the developers so that RPA can be successfully
deployed in the long run. Management also needs to properly identify the tasks that can be automated,
appoint a team champion, and create process automation policies. The team champion ensures that
appropriate RPA technologies are used for automation.
Standardized machine-readable documents for RPA The accounting firm must standardize the documents that are used for RPA inputs. Machine-readable
input inputs are critical to making RPA work.
Attended RPA is more suitable for most audit tasks RPA may not ideally automate and may not be suitable for every audit task. Some audit tasks require
compared to unattended RPA specialized auditors’ judgement, are highly sophisticated and complex, and therefore are not suitable for
RPA. Attended RPA can have more widespread applications in audit, as compared to that of unattended
RPA. It is essential overall for accounting firms to have access to an on-site RPA team during the
implementation process so as to promptly provide the firm with the necessary support when needed.
Since RPA is a relatively new concept, the lack of experience in this area can portend an extended
implementation period.
Employees’ familiarity with RPA Some current employees may not be familiar with RPA as yet and thus may not be equipped with the
technical knowledge to adequately perform RPA functions. Therefore, the introduction of RPA may face
pushbacks, particularly when compounded with the fact that auditors often have to cope with tight
deadlines from multiple audit engagements.
Collaborations with clients The clients may be at different stages of digital transformation, and not all of them may have digitized
their business tasks. RPA thus requires close collaborations between the accounting firms and their
clients.
RPA and cybersecurity issues Personal and corporate data must be well protected within the RPA systems. As technology advances,
cyberattacks are becoming more frequent and sophisticated. Hence, to avoid security breaches, it is
critical that accounting firms remain vigilant and continually update their RPA systems through security
patches.

the previous year’s stat audits, were not yet automated in the prototype we developed (see Fig. 14 in the Appendix). Here, the RPA bot
was developed to help auditors request documents, sort the documents, and extract some details from the documents for further
processing. RPA can help auditors save half a day of work. This time saving is due to the bot’s ability to quickly extract relevant data
from customers’ documents. The auditor then further evaluates this information. The auditors felt that the bot is helpful for them
because it allows them to expeditiously access relevant information for use in evaluating the stat audit. The development of this RPA
bot was relatively simple, as it only involved requesting and extracting documents. We were able to deploy a bot that was well-versed
in the document details so as to extract the relevant data. Once the data was extracted, reviewers would need to verify the accuracy of
the data, especially in terms of dates and numbers. In addition to the “document-literate” bot, we connected the RPA bot to a
spreadsheet to automatically input the data in an appropriate format. We did not use any extensive formulas in the spreadsheet. A total
of two months is estimated for the development of this prototype, as presented in Table 6.

4. Discussion: Future RPA challenges and opportunities

Accounting firms are constantly facing fee and time pressures arising from a multitude of client and regulatory expectations. This is
especially so during the COVID-19 pandemic period, when many companies are put to the tests. With the help of technologies such as
RPA, the accounting industry can evolve and transform the way traditional audits are carried out, thereby opening up new avenues for
greater effectiveness and efficiency. Proper implementation of RPA can free up time for auditors to address the issues that matter most
to their clients. However, as we will elaborate, RPA can present a myriad of challenges and opportunities in the road ahead. Table 7
summarizes the practical implications which we have drawn from our RPA development journey. These practical implications
complement the five key themes related to workforce, IT governance, privacy and security, system sustainability, and the measurement
of RPA success that Zhang et al. (2022) have previously identified as important for companies to consider when adopting RPA in the
accounting domains.
Some of our current practical implications reinforced the key themes suggested by Zhang et al. (2022). First, similar to the
importance of IT governance in the fields of Development and Operations (DevOps) as well as cloud computing (Plant et al., 2022;
Prasad et al., 2014; Prasad and Green, 2015), IT governance also matters in RPA implementation. We find that governance in the
latter’s case can be established with proper communication, particularly when aided by the presence of an RPA champion. The
champion can help to facilitate proper application of business process management lifecycle for automation implementation and can
also help to ensure that the bots are deployed on schedule so as to maintain the system’s long-term sustainability. While our

11
A. Perdana et al. International Journal of Accounting Information Systems 51 (2023) 100641

decentralized RPA design can result in a quick turnaround to bot deployment, a lack of transparent governance can increase the level of
risk when automation is expanded and deployed on a larger scale. Effective RPA governance is thus needed to manage both the
complexity in implementation and also the ambiguity in business process automation.
Second, data security issues may arise due to poor business processes and/or deficient bot designs. From our observations, com­
panies must continue to stay vigilant of their data governance and ensure that they have sufficient security firewalls to effectively
protect their RPA systems. Data governance should hence be embedded within their IT governance initiative. In doing so, one’s data
can be readily and securely managed by RPA. As the prevailing system of checklists and proposals used for RPA governance are mostly
piecemeal, more extensive research will need to be conducted in this area, so as to provide more empirical evidence of what constitutes
rigorous and uniformly effective industry policies toward RPA governance. Table 7 presents the summary of our practical findings.

4.1. Future RPA challenges

Currently, some audit processes require substantial amount of manual work, such as document matching, reconciliation, and re­
view, which can in turn reduce the overall audit productivity (Bakarich and O’Brien, 2021). In general, strategic implementation of
RPA requires a concerted organizational commitment to make and direct the changes (Lacity and Wilcocks, 2021). Although RPA can
significantly improve the effectiveness and productivity of the audit processes, one should consider the challenges that may arise from
implementing RPA.
In the cases we presented, we used the free RPA software as part of our academic engagement to show students how to apply RPA in
the real world. Therefore, minimal cost was involved in developing the automation prototype. If accounting firms wish to deploy the
automation solutions on a larger scale over a prolonged period, they will likely need to consider the cost of procuring more sophis­
ticated software packages. Overall, the initial challenges are the start-up and maintenance costs (Cooper et al., 2019; Lacity and
Wilcocks, 2021). The former costs relate to setting up RPA, such as purchasing software, redesigning the automation process, and
training the staffs to operate RPA. The latter costs include the proper deployment and operation of RPA within the accounting firms.
For example, in scenario 4, the auditors will favor automating this process because it allows them to devote more time on essential
matters and less time on administrative tasks. Automation may be desirable but not economically feasible from a firm’s perspective, as
implementing RPA can be expensive in the initial stages. The accounting firm in scenario 4 needs to identify all tangible costs and
benefits, including those one-time and recurring costs. An example of recurring costs is the training costs incurred for both the ac­
counting and client firms’ personnel in scenario 4. Hence, in varying degrees, both parties will likely have to bear such recurring costs.
By implementing costly audit processes, it also often results in an increase to the audit fees. As a result, the accounting firm in scenario
4 may inadvertently put itself at a competitive disadvantage, given the aggressive pricing environment and low switching costs in some
audit sectors. Therefore, it is crucial to weigh such impacts when deciding if and how much to automate the various processes.
Before the firm implements RPA to automate its processes, management must assess the current level of automation and determine
how long it will take the entire firm to use the technology efficiently. Management needs to properly identify the tasks that can be
automated, appoint a team champion, and create process automation policies. The team champion ensures that the appropriate RPA
technologies are used for automation (Perdana and Yong, 2021). In the course of implementation, management must also review the
mechanisms used to monitor the development process and assess the effectiveness of RPA for specific industries and customers. After
the implementation, auditors can then brainstorm to determine if any problems or errors have occurred, and whether these issues have
been promptly resolved.
Moreover, unlike traditional approaches, the continued use of RPA may require substantial maintenance costs, even if the processes
are highly repetitive (Kokina and Blanchette, 2019). Therefore, it is recommended that the accounting firm budgets adequately for
maintenance when implementing RPA. In general, maintenance is needed when three broad issues arise: (1) automation script change,
(2) insufficient IT infrastructure to support the RPA application, and (3) problems with the RPA tools. Good budget formulation is
imperative because without a sufficient budget for maintenance, the company can risk breaching confidentiality in the event when
information is leaked, which can in turn distract from the initial RPA objectives. The amount invested in maintaining RPA also involves
training employees to quickly adapt to changes. As operational and accounting standards change from time to time, it can be a constant
challenge to keep up with RPA maintenance, especially with different departments’ vested interests and expectations toward RPA. RPA
developers thus need to be in constant communications with all affected departments, with the latter also needing to provide periodic
performance feedback to the developers. Such mutual cooperation will ensure that RPA can be successfully deployed in the long run.
Second, the accounting firm must standardize the documents that are used for RPA inputs (Huang and Vasarhelyi, 2019). Machine-
readable inputs are critical in making RPA work. Although there is an evolving movement toward a paperless society (Campbell,
2020), we found that during our prototyping, there is still a significant number of clients who are hesitant to digitize their documents.
Since RPA is best suited for consistent and standardized data, this mode can present a problem because documents are often delivered
from various sources in practice. Their format may vary depending on the nature of the transactions and the terminologies used. For
example, when reviewing data, two of the same data may be reported in different formats (for e.g., “sales” may be recorded as
“revenue”), thus resulting in some items being unnecessarily flagged as false positives. Inconsistencies can affect the interpretation of
data attributes, hence presenting a challenge during RPA implementation. Further, digitizing the paper-based documents could create
problems, as mentioned in scenarios 2 and 3. Scanned documents may end up distorting the numbers and texts. Hence, the OCR tool
within RPA may sometimes produce inaccurate results when extracting numbers and texts from the scanned documents.
In addition, companies may use different accounting and invoicing systems, thus the numbering systems or the formatting of the
invoice dates may vary. It is, therefore, almost impossible to develop an “all-inclusive” RPA process. The accounting firm may need to
invest significant time, money and effort in building an RPA team that can develop an automated system specific to their client’s

12
A. Perdana et al. International Journal of Accounting Information Systems 51 (2023) 100641

business processes and accounting software (Perdana and Arisandi, 2021). This initiative also requires auditors’ involvement in
creating workflow diagrams, specific business rules and feedback on implementation. The accounting firm must also ensure that
changes made through the RPA process are immediately reflected in shared documents. Therefore, it is essential to have access to an
on-site RPA team during the implementation process, so as to promptly provide the firm with necessary support when needed. Since
RPA is relatively new to the market, the lack of experience in this area can result in an extended implementation period.
Third, RPA may not perfectly automate and hence may not be suitable for every audit task (Moffitt et al., 2018). Some audit tasks
require more specialized auditors’ judgement, are highly sophisticated, and therefore are not suitable for RPA. In contrast, some audit
tasks require fewer auditors’ judgement. For example, the ELT at AFA is more mechanical and can require more manual recalculation
and cross-referencing. In this case, attended RPA can become more applicable, and auditors can then spend the additional time on
more critical tasks. In addition, each organization has its business processes organized to fit the nature of the business. This can pose a
challenge for RPA development, since bots are currently not capable of being so comprehensive as to cover all possible business
processes. RPA components are generally applicable only to specific tasks. This may lead to a lack of adaptability or cross-usability of
RPA components (Kedziora and Penttinen, 2021). For example, even if there is an RPA for ELT, some of the RPA components cannot be
used for other audit work, such as for bank reconciliation, because both the data and nature of the tasks are different. Ultimately, this
can pose a challenge in executing an effective RPA for audit clients across different industries.
Fourth, there is a possibility that current employees may not be familiar with RPA as yet and thus may not be equipped with the
technical knowledge to perform RPA functions (Kokina and Blanchette, 2019). The introduction of RPA may therefore face pushbacks,
particularly when compounded with the fact that auditors often have to cope with tight deadlines from multiple audit engagements.
Further, any misguided information about disruptive technologies and their impact can cause further anxiety, as auditors naturally fear
that they may lose their jobs once robots take over their job responsibilities in the future (Cooper et al., 2019). To allay such anxiety,
proper communication channels between the accounting firm and its employees need to be established, to ensure that employees better
understand how RPA serves to help, rather than (the misguided notion) to replace, them (Perdana and Arisandi, 2021). It will be hard
to get buy-in from the employees if they do not understand that RPA provides a win–win solution to improve processes. After
implementing RPA for a particular audit process, auditors need to know how it works and how to use it effectively for their tasks. The
accounting firm should thus organize a company-wide training program to familiarize auditors with RPA and to adapt it to each
client’s business, notwithstanding that this move can portend the firm incurring higher costs for employees’ RPA training.
Fifth, RPA requires close collaboration between the accounting firms and their clients. Not all of the clients have already suc­
cessfully digitized their business tasks. The clients may be at different stages of digital transformation. For example, some clients may
still be using manual processes and are still keeping paper documents, instead of having already transitioned to digital copies. Small
and medium-sized enterprises, in particular, may not have sufficient resources to digitize their business tasks. If the clients have not
fully transitioned to digitization, it is difficult for RPA bots to be successfully deployed during the audit. Some clients may also not have
digital or scanned copies of all their payment documents or vendor invoices during the AFB’s SFUL processes. As a result, some audit
tasks cannot be automated with RPA, such as that concerning notification when thresholds exceed SAD or when the need arises to
capture data from multiple document sources. Another area of concern is in the automation of stat audits, which can also put audit
firms in a bind. By adopting RPA in auditing, firms engaging secretarial services may need to sign up for the AFD audit’s application as
well as the audit platforms of other accounting firms, to facilitate uploading of all the documents onto the audit portal. With confi­
dentiality issue as a potential impediment, firms may sometimes be reluctant to get on board, which will in turn delay the adoption
process.
Sixth, there has been an ongoing debate on the need for organizations to address issues concerning RPA cybersecurity and privacy
(Syed et al., 2020). Surveys that were conducted with professionals from the Big 4 accounting firms have suggested that clients can be
reluctant to use RPA due to concerns about data privacy as well as the need to protect business processes and workflows (Cooper et al.,
2019). One challenge that accounting firms face after implementing RPA is the need to continuously strengthen cybersecurity (Moffitt
et al., 2018). For example, every time the RPA bots send clients’ emails to request documents, the usernames and passwords are stored
in the RPA. Similarly, the digitized documents that are submitted by clients are stored in encrypted cloud storage systems, with the
usernames and passwords again stored in the RPA. While these passwords are typically encrypted and securely stored, the accounting
firms’ personnel will need to continually monitor the access logs within these RPA bots.
Although RPA can automate the detection of cyber threats and defend against malware attacks, hackers have learned to use AI and
ML to penetrate all types of security systems.1 Cybercriminals can rely on ML to understand an organization’s network and develop
malware to attack its network. Accounting firms need to invest more in their cybersecurity systems to ensure that a data breach will not
end up compromising the corporate network. Any data loss can have serious consequences, particularly in the case of banks that hold a
large amount of confidential information. Because of this, banks do not typically send information directly to auditors for fear of
malicious users, and their files are additionally protected with passwords. A data breach can lead to the disclosure of sensitive in­
formation concerning other banks, like the fee rates set in the engagement or invitation letters. Therefore, personal and corporate data
must be well protected within the RPA system.
As technology advances, in order to avoid security breaches from increasingly sophisticated cyberattacks, it becomes critical for
accounting firms to remain vigilant and to continually update their RPA systems through security patches. Before implementing
automation, precautions (for e.g., using encryption to limit unauthorized access to customer data) should be taken to ensure data

1
https://www.cio.com/article/189253/boosting-security-with-robotic-process-automation-2.html.

13
A. Perdana et al. International Journal of Accounting Information Systems 51 (2023) 100641

Fig. 1. As-is Business Process for Employee.

privacy and confidentiality.

4.2. Future RPA opportunities

Despite the challenges of RPA, its benefits are also clearly evident. A survey conducted by Deloitte in 2018 found that 90% of
respondents saw success in the quality and accuracy of their work after implementing RPA (Wright et al., 2018). In addition, 86% of
respondents said productivity had improved in their organization, and 59% agreed that they had achieved cost savings through RPA. In
terms of compliance, 92% of respondents felt that compliance improved after implementing RPA within their business tasks.
From our experiences and conversations with both the students and firms’ personnel over the course of the RPA implementation,
we believe that, other than the time involved in the RPA conceptualization and development, the key costs mainly revolved around the
process switchover, i.e., the additional time taken to train the students and firms’ personnel in adopting the RPA bots, a process that
often involved learning new procedures and protocols, as well as the time required for all parties involved to periodically monitor the
implementation progress. With regard to the UiPath software that we had used to develop automation solutions in our RPA imple­
mentation, the license for use then was free of charge. In addition to UiPath’s standard RPA features, during the time when we
developed the automation solutions (from 2019 to 2021), UiPath was allowing free access to its intelligent automation capabilities (a.
k.a., A.I. Fabric) program. Therefore, there was no quantifiable software costs in this regard. Also, as our consultancy work and the
students’ industry work engagement programs were all within the scope of the academic programs concerned, the costs (not
considering opportunity costs) were presumably nominal. However, any commercial enterprise contemplating long-term RPA
engagement will realistically need to take into consideration the potential for substantive outlay in this regard.
As for the benefits, the most substantive value comes from the estimated time savings in those RPA scenarios involved, hence
portending longer-term effectiveness and/or efficiency gains in the various audit procedures if continued to be implemented. Overall,
based on our anecdotal experience interacting with the personnel and senior audit teams during the course of the various RPA
implementations, all of the four firms involved have felt that the time invested toward learning RPA was definitely worthwhile, with
the benefits (particularly long-term savings) clearly exceeding the costs. Hence, our findings mainly signify a successful first step in the
direction of advocating wider-scale RPA implementation for more wide-ranging audit scenarios. Notwithstanding our (and the firms’)
observations of a more streamlined and efficient process in the various areas of our RPA implementations, with clearly less human-led
procedural inconsistencies (in both aspects of time and process), our study is admittedly limited by precise, quantified information
with regard to actual reduction of errors.

14
A. Perdana et al. International Journal of Accounting Information Systems 51 (2023) 100641

Fig. 2. To-be Business Process for Employee Loan Testing.

4.3. Employees’ perspectives

A survey conducted by Persellin et al., (2019) indicated that auditors generally work very long hours. During peak seasons, the
weekly work hours can average 65 to even 80 h (Persellin et al., 2019). This can be attributed to the staffing shortage issue facing many
accounting firms, which has led to many accountants burning out and even quitting. This problem has resulted in many workload-
related audit deficiencies. Contrary to a misconstrued notion that automation means a machine or robot taking over what is typi­
cally done by humans, the reality is that in most cases, automation does not take work away from humans. Instead, it simplifies a
multitude of activities, leaving the accountants to complete the more value-added work. In the long term, automation helps to lower
operating costs, thereby increasing a company’s competitiveness and performance (Asatiani and Penttinen, 2016; Lacity and Wilcocks,
2021).
By implementing RPA, auditors can also save time on their engagements since it eliminates the painstaking need of having to
physically reconcile documents (Huang and Vasarhelyi, 2019). As a result, more time can be spent focusing on higher-risk audit tasks.
This can increase auditors’ productivity and efficiency on their assignments, thereby reducing the likelihood of having to work
overtime and the associated turnover problem. In addition, human errors can be minimized, which will positively impact productivity
and efficiency (Syed et al., 2020). In the end, improved efficiency results when all the man-hours normally required to perform an
analysis can be completed by an automated test application in a much shorter amount of time, thus saving valuable labor hours.
Eliminating labor hours while maintaining the quality of work is not only cost-saving but also process-efficient. This is evident in
our prototyping where automation can save auditors’ time in conducting the audit processes. Work appraisal wise, this time saving can
in turn help to improve the auditors’ key performance indicators (KPI). Undeniably, introduction of automation in the workplace will
result in increased effectiveness and efficiency (Kedziora and Penttinen, 2021).
Further, the synergistic interaction between human and RPA can often lead to human–AI augmentation (Rinta-Kahila et al., 2021).
One will not have to manually enter and analyze information unless anomalies are detected. In addition, employees who are auto­
matically notified of a problem can take corrective action instantaneously, instead of having to manually communicate back and forth
amongst the different entities. This leaves more time for individualized decision-making, based on data that can be further collected.
By standardizing workflows through automation, the organization can thus focus on critical tasks and employees can then be more

15
A. Perdana et al. International Journal of Accounting Information Systems 51 (2023) 100641
Fig. 3. As-is Business Process for Bank Reconciliation.
16
A. Perdana et al. International Journal of Accounting Information Systems 51 (2023) 100641
Fig. 4. To-be Business Process for Bank Reconciliation.
17
A. Perdana et al. International Journal of Accounting Information Systems 51 (2023) 100641

Fig. 5. As-is Business Process for Analytical Procedure.

Fig. 6. To-be Business Process for Analytical Procedure.

Fig. 7. As-is Business Process for Sales.

18
A. Perdana et al. International Journal of Accounting Information Systems 51 (2023) 100641

Fig. 8. To-be Business Process for Sales.

Fig. 9. As-is Business Process for Purchase.

fruitfully deployed to more value-added tasks involving problem-solving and ideas-developing. This will empower the employees,
provide a more rewarding work experience, and ultimately contribute to the company’s bottom line (Lacity and Wilcocks, 2021).
Finally, the use of RPA can enable auditors to learn new value-added skills. With the advent of disruptive technologies, many
existing jobs will be transformed as automation takes hold (Frey and Osborne, 2017). With RPA and the need for ongoing education to
keep up with new opportunities that may arise from automation (for e.g., audit analytics), a new era for the accounting profession
awaits.

4.4. Employers’ perspectives

Accounting firms that use RPA are likely to see savings in their labor costs (Moffitt et al., 2018). After all, organizations should view
the automation process as a strategic investment rather than as a sunk cost (Lacity and Wilcocks, 2021). Even though the initial
implementation costs can be costly, automation can prove to be a worthwhile long-term investment once those costs are recovered,
thus eliminating the need for excessive paperwork, extensive labor, painstaking maintenance and high repair costs (Cooper et al.,
2019). Firms can then use these savings to invest in more sophisticated automation programs that will further benefit their employees.
RPA can also be used to improve the efficiency of process workflows, with the employees’ effort redirected to revenue-generating

19
A. Perdana et al. International Journal of Accounting Information Systems 51 (2023) 100641

Fig. 10. To-be Business Process for Purchase.

Fig. 11. As-is Business Process for Search for Unrecorded Liabilities.

activities and other essential tasks. As accounting firms move toward digitalization, thereby embracing the digital storage of working
papers and clients’ information, it has become increasingly important for accounting firms to take data security seriously (Cooper
et al., 2019). Further, with the decentralization in information collection, audit documentation and process logs can present a problem
for internal auditors. RPA can be of help here by enabling a centralized collection of information, thereby limiting who and the fre­
quency with which such sensitive data can be accessed. This will mean that only the IT department has permission to manage the bot.
This initiative can thus streamline auditability, security and also help address governance issues by limiting the use of unauthorized
software, latter often leading to data leaks.
Ultimately, the benefits of automation to a business will become apparent as more companies adopt this line of new technologies.
More output can be produced for less money, resulting in cost savings and improved accuracy. It also allows for better control and
visibility of processes and allows employees to do their jobs more efficiently and effectively. By automating workflows and tasks,
companies can thus benefit from the return on investment that RPA offers.

5. Conclusion

In this paper, we present four scenarios from the perspectives of four accounting firms. For each scenario, we explain the business

20
A. Perdana et al. International Journal of Accounting Information Systems 51 (2023) 100641

Fig. 12. To-be Business Process for Search for Unrecorded Liabilities.

Fig. 13. As-is Business Process for Search for Stat Audit.

tasks and processes that can potentially be automated. We then present the possible automation strategies, and also analyze the
challenges and opportunities of implementing automation in these audit practices. Evidently, RPA adds value for both the firms and
clients. More importantly, it helps make auditing a sustainable career for the firms’ employees. Our study, however, has some limi­
tations. First, our prototype solutions are decentralized, limited to only specific business processes, and they are disconnected from
other business processes. Therefore, our prototype may not fully show how RPA can support the entire audit processes. Second, our
study is admittedly limited by precise, quantified KPI information with regard to actual reduction of errors. This results in our inability
to compare the error rates before automation with the error rates after automation.
Future research can hopefully extend this study and investigate more quantified data in regard to the actual costs and benefits of
RPA implementation over a longer time period from the developmental to maintenance phases, using more rigorous data collection
and testing techniques, and also potentially taking into consideration differing firm characteristics like firm size, location, regulatory
standards and even personnel dynamics and/or cultural predispositions. RPA has just begun to gain a foothold in the audit industry.
Ultimately, professional judgment is still the cornerstone of audit practices. Looking forward, auditors must prepare for a pivot toward
automation, algorithms and analytics. What auditors will need to succeed is to remain intuitive, apply their professional judgment, and
take an exploratory approach toward cutting-edge audit procedures and technologies, so as to continue making a difference in adding
value for their clients.

21
A. Perdana et al. International Journal of Accounting Information Systems 51 (2023) 100641

Fig. 14. To-be Business Process for Search for Stat Audit.

CRediT authorship contribution statement

Arif Perdana: Conceptualization, Formal analysis, Investigation, Writing – original draft, Writing – review & editing. W. Eric Lee:
Formal analysis, Writing – review & editing. Chu Mui Kim: Conceptualization, Formal analysis, Investigation, Writing – review &
editing.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to
influence the work reported in this paper.

Data availability

The data that has been used is confidential.

Appendix

References

Asatiani, A., Penttinen, E., 2016. Turning robotic process automation into commercial success – Case OpusCapita. J. Inform. Technol. Teaching Cases 6, 67–74.
https://doi.org/10.1057/jittc.2016.5.
Bakarich, K.M., O’Brien, P.E., 2021. The Robots are Coming But Aren’t Here Yet: The Use of Artificial Intelligence Technologies in the Public Accounting Profession.
J. Emerging Technol. Account. 18 (1), 27–43.
Campbell, A., 2020. Digital Transformation Beats Paperless Processes. CPA Practice Advisor.
Cooper, L.A., Holderness, D.K., Sorensen, T.L., Wood, D.A., 2019. Robotic Process Automation in Public Accounting. Account. Horiz. 33 (4), 15–35.
Eulerich, M., Pawlowski, J., Waddoups, N., Wood, D., 2022. A Framework for Using Robotic Process Automation for Audit Tasks. Contemp. Account. Res. 39,
691–720. https://doi.org/10.1111/1911-3846.12723.

22
A. Perdana et al. International Journal of Accounting Information Systems 51 (2023) 100641

Frey, C., Osborne, M., 2017. The future of employment: How susceptible are jobs to computerisation? Technol. Forecasting Social Change 114, 254–280. https://doi.
org/10.1016/j.techfore.2016.08.019.
Huang, F., Vasarhelyi, M., 2019. Applying robotic process automation (RPA) in auditing: A framework. Int. J. Account. Inf. Syst. 35, 100433 https://doi.org/10.1016/
j.accinf.2019.100433.
Kedziora, D., Penttinen, E., 2021. Governance models for robotic process automation: The case of Nordea Bank. Journal of Information Technology Teaching Cases 11,
20–29. https://doi.org/10.1177/2043886920937022.
Kokina, J., Blanchette, S., 2019. Early evidence of digital labor in accounting: Innovation with Robotic Process Automation. Int. J. Account. Inf. Syst. 35, 100431
https://doi.org/10.1016/j.accinf.2019.100431.
Lacity, M., Wilcocks, L., 2021. Becoming Strategic with Intelligent Automation. MIS Q. Exec. 20, 1–14.
Leung, P., Coram, P., Cooper, B., Richardson, P., 2015. Modern Auditing and Assurance Services, 6th Edition. John Wiley & Sons Australia.
Moffitt, K.C., Rozario, A.M., Vasarhelyi, M.A., 2018. Robotic Process Automation for Auditing. J. Emerging Technol. Account. 15 (1), 1–10.
Perdana, A., Arisandi, D., 2021. Automation at Truveil: Business Process Understanding and Analysis in RPA Implementation. J. Emerging Technol. Account. 19 (2),
141–154.
Perdana, A., Yong, K., 2021. Intelligent Automation. ISCA Journal 39–41.
Persellin, J.S., Schmidt, J.J., Vandervelde, S.D., Wilkins, M.S., 2019. Auditor Perceptions of Audit Workloads, Audit Quality, and Job Satisfaction. Account. Horiz. 33
(4), 95–117.
Plant, O., van Hillegersberg, J., Aldea, A., 2022. Rethinking IT governance: Designing a framework for mitigating risk and fostering internal control in a DevOps
environment. Int. J. Account. Inf. Syst. 45, 100560 https://doi.org/10.1016/j.accinf.2022.100560.
Plattfaut, R., Borghoff, V., 2022. Robotic Process Automation – A Literature-Based Research Agenda. J. Inf. Syst. 36 (2), 173–191.
Prasad, A., Green, P., 2015. Governing cloud computing services: Reconsideration of IT governance structures. Int. J. Account. Inf. Syst. 19, 45–58. https://doi.org/
10.1016/j.accinf.2015.11.004.
Prasad, A., Green, P., Heales, J., 2014. On governance structures for the cloud computing services and assessing their effectiveness. Int. J. Account. Inf. Syst. 15,
335–356. https://doi.org/10.1016/j.accinf.2014.05.005.
Rinta-Kahila, T., Penttinen, E., Lyytinen, K., 2021. Organizational transformation with intelligent automation: Case Nokia Software. J. Inform. Technol. Teaching
Cases 11, 101–109. https://doi.org/10.1177/2043886920954874.
Schatsky, D., Muraskin, C., 2016. Robotic process automation A path to the cognitive enterprise [WWW Document]. Deloitte Insights.
Syed, R., Suriadi, S., Adams, M., Bandara, W., Leemans, S., Ouyang, C., Hofstede, A., van de Weerd, I., Wynn, M., Reijers, H., 2020. Robotic Process Automation:
Contemporary themes and challenges. Comput. Ind. 115, 103162 https://doi.org/10.1016/j.compind.2019.103162.
Thibodeau, J., Strawser, J., Louwers, T., Sinason, D., Blay, A., 2017. Auditing & Assurance Services. McGraw Hill.
Wilkin, C.L., Chenhall, R.H., 2020. Information Technology Governance: Reflections on the Past and Future Directions. J. Inf. Syst. 34 (2), 257–292.
Wright, D., Witherick, D., Gordeeva, M., 2018. The robots are ready. Are you? Untapped advantage in your digital workforce.
Zhang, C.(., 2019. Intelligent Process Automation in Audit. J. Emerging Technol. Account. 16 (2), 69–88.
Zhang, C., Thomas, C., Vasarhelyi, M.A., 2022. Attended Process Automation in Audit: A Framework and A Demonstration. J. Inf. Syst. 36 (2), 101–124.
Zhang, C., Issa, H., Rozario, A., Soegaard, J.S., 2023. Robotic Process Automation (RPA) Implementation Case Studies in Accounting: A Beginning to End Perspective.
Account. Horiz. 37 (1), 193–217.

Arif Perdana works as Associate Professor in Data Science at Monash University. Prior to his academic career, Arif worked in the IT service and finance industry. Arif’s
diverse research illuminates the transformative role of digital technologies such as algorithmic systems, automated decision-making, data analytics, and blockchain in
various sectors, from finance to education to healthcare transformation. He has provided critical insights into the challenges of adopting digital technologies while
exploring technology-driven strategies across sectors. Arif has conducted industrial research and consulting with institutions and companies in Singapore and Indonesia.
His research has been published in Information & Management, Electronic Markets, Behaviour & Information Technology, the International Journal of Accounting Information
Systems, Journal of Information Systems, Telematics and Informatics, Technology in Society, Australasian Journal of Information Systems, and Journal of Information Technology
Teaching Cases. Contact Address: Email: [email protected]. Monash University, Indonesia.

W. Eric Lee is an associate professor and the accounting alumni faculty fellow in the accounting department of the University of Northern Iowa. His previously
published and ongoing research examine: the behavioral nuances in implementing corporate sustainability accounting, the narrative effects of media releases con­
cerning blockchain/ fintech and corporate social responsibility reporting, the various pedagogical techniques in fraud and forensic accounting, issues of ethics and
community awareness among accounting students, and perception toward cryptocurrencies and other investments. His empirical research has previously been pub­
lished in the Journal of Business Ethics, Advances in Accounting, Telematics and Informatics, Business and Professional Communication Quarterly, Social and Environmental
Accountability Journal, Advances in Accounting Education, Journal of Forensic and Investigative Accounting, amongst others. He has previously served as the sectional chair/
liaison and his research had been presented in the American Accounting Association’s national, regional, and sectional conferences, the Conference of the Association for
Business Communication, as well as the European Business Ethics Network Conference. He has previously worked in both public and corporate accounting, and is a
licensed CPA, CGMA, CITP and CFE.

Chu Mui Kim Mui Kim graduated with an accountancy degree from NUS with second-class honours. She is a Chartered Accountant of Singapore and is a member of the
Institute of Singapore Chartered Accountants. She also has a Master’s degree in Business Administration from the University of Southern Queensland. Mui Kim has vast
experience in teaching auditing, financial accounting and management accounting related modules at tertiary and professional qualification levels and has been the
recipient of various teaching excellence awards by the institutions of higher learning. She is also an effective trainer in the area of financial training for corporations and
has participated in consultancy projects involving business process reengineering and special audit reviews. As a Senior Lecturer with the Singapore Institute of
Technology, Mui Kim has been awarded research grants and successfully completed projects that help small and medium size audit firms to automate some of their
practices. She has also received teaching and learning grants to develop innovative ways of auditing education. Mui Kim’s main research interests include innovations in
auditing and accounting and corporate governance and have published works on these areas.

23

View publication stats