ICAEW.com works better with JavaScript enabled.

The essential guide to robotic process automation

December 2020
Exclusive content
Access to our exclusive resources is for specific groups of students and members.

Robotic process automation (RPA) can save organisations' time, with tasks performed with higher speed and accuracy, as well as removing the risk of human error. However, it is not suitable for all processes and, in some cases, could act as a sticking plaster for a system that needs to be re-engineered.

What is RPA?

RPA is software that automates tasks by working in parallel with applications already used by businesses to extract or process data. This could be accounting, customer relationship management or enterprise resource planning software, with processes typically undertaken by humans being replaced by RPA.

Within the context of RPA, the terms robot and robotic are often misnomers as they conjure up images of physical robots more commonly associated with the manufacturing of physical goods. However, RPA robots are primarily concerned with automating processes related to software usage, not the physical creation or modification of products.

RPA automates existing tasks, as opposed to creating processes from scratch. This includes everyday manual tasks such as copying and pasting text, retrieving documents from emails and entering information into a database.

The types of tasks undertaken by RPA are not particularly complex, with software robots’ rote following a pre-defined sequence or list of instructions. For Excel users, a relevant analogy is that they can similarly recreate the keystrokes used to program macros.

RPA is often considered as an alternative to outsourcing or offshoring as it shares many of the same benefits but does not require as much supply chain management or create as many concerns around access and security.

Automating repetitive rule-based tasks

“Business cases for RPA should come from processes that have high volumes of transactions, are relatively simple and can have clear rules applied to them. The best-suited processes are those which require manual intervention but are not value-adding to organisations,” says Kirstin Gillon, Technical Manager in ICAEW's Tech Faculty. 

For accountants working in industry, this is likely to include tasks such as reconciliations and month-end journals (accruals, prepayments and depreciation) and management of accounts payable and receivable. Tasks in financial services-led companies suitable for RPA include credit checks, claim processing and fraud detection.

As these tasks follow a clear set of instructions (“if this, then that” logic), applying RPA will allow them to be completed faster and more accurately than by humans. Unlike humans, robots don’t need a break, so conceivably can continuously process tasks 24 hours a day. Assuming robots have been set up correctly, accuracy measures, related to data extraction and reconciliations, will also mitigate the risk of non-compliance and fines from the tax authority and sector-specific regulators.

Additionally, using RPA to complete these tasks will free up staff to be more productive, including by working on less process-led and more value-adding assignments.

Unlike other types of automation solutions, businesses can benefit from RPA relatively quickly as processes are not being reinvented. This also means that, in most instances, minimal training is required for staff. Many RPA vendors have a drag and drop interface, allowing staff to easily and intuitively map out sequences of commands that they are already familiar with.

Deploying RPA does not require programming skills, but it is important to include IT departments in related projects to manage issues around governance and security risks. Additionally, discussions between IT and finance around deployments will enhance the effectiveness of integrations, as well as creating a consensus on who is responsible for different elements.

Is RPA relevant for your organisation?

While RPA can have clear benefits, it will not be suitable for every organisation. For example, consideration will need to be given to whether robots are appropriate for customer-facing interactions. Unlike humans, RPA cannot display empathy, so automated responses to customer queries may strike the wrong chord. This could result in a negative view of a company based on a complaint that requires a non-standard answer. Therefore, during the vendor selection process, companies should ask how the RPA application will handle exceptions to standardised questions and responses. 

Additionally, a criticism of RPA is that it is a short-term solution for processes and systems that require re-engineering from the ground up.

“RPA is to some extent a sticking plaster. In many instances, RPA automates processes that are not very well designed in the first place. In certain cases it can be useful, but in others, the right solution for companies may be to re-engineer processes and reintegrate systems to improve efficiency further,” Gillon says.

Cost and size

Publicly available data from vendors indicates that licences for robots are several thousand pounds a year, but other associated costs tend to be priced on a bespoke basis, taking into consideration processes and complexity.

“The pricing structure of vendors varies, which makes comparability difficult," says Rick Payne, Finance Direction Lead at ICAEW's Business and Management Faculty. "Related costs often include licences per robot (based on volume usage), consultancy support during set up and infrastructure set-up costs.”

As a rule of thumb, from an output and cost perspective, companies should seek a return on investment from a successful RPA deployment replacing the output of three equivalent full-time staff.

However, RPA can generate successful results for companies of all sizes. Dave White, CEO at business advisory practice White Bruce, has run a number of RPA projects for SME clients to improve the efficiency of their operations. He believes that the adoption of cloud accounting software has made it easier for SMEs to deploy RPA relatively easily, something that previously would have been the preserve of companies who could afford expensive consultants.

White says: “The big change came with cloud migration. Suddenly tiny data sets in companies held in their broom cupboards became linked. More accessible technologies, such as webhooks (which connect two different services together without the need for Application Program Interface (API) knowledge) and Zapier are taking automation to the next level for small businesses."

Zapier is an online automation tool that connects apps and web services without needing to build third-party integration. One of the more popular integrations of Zapier with cloud accounting software is a connection that automatically creates new invoices on Xero related to Paypal sales.

There are a number of different RPA vendors on the market, most of which publish research papers that can be handy for assessing similar projects.

With the market changing rapidly, companies should be mindful of vendors rebadging their capabilities as RPA when they are actually providing an existing automation software technology. Giveaways include long-standing features with descriptions related to having RPA functionality. Similarly, some vendors will list artificial intelligence (AI) capabilities but in practice this is rarely the case. That said, it is possible to augment RPA with third-party AI technology to enhance automation further.

Businesses considering RPA for the first time may choose to incorporate a robot to complete one specific task first and, if successful, incorporate further deployments. Some robots could be used on more than one task or require basic modifications.

Should we tax robots?

Advances in technology over the last decade or so have made it clear that many manual tasks conducted in the workplace can be automated.

Research from PwC in 2017 shows that 45% of all work activity could be automated, reducing global labour costs by $2trn. Even if this estimate is only marginally accurate, this will result in governments taking a significant hit from the revenues they generate from employment taxes, which in the UK make up around 44% of all collected taxes.

This has resulted in policymakers, and even some business leaders, suggesting that we should tax robots to replace lost revenues. Bill Gates, founder of Microsoft, told business news site Quartz in 2017 that “if a robot comes in to do the same thing, you’d think we’d tax the robot at a similar level”.

However, opinions vary and highlight a wider problem to do with public funding. In a January 2018 City Am editorial, Frank Haskew, Head of ICAEW’s Tax Faculty, wrote that a robot tax “looks to be a short-sighted, knee-jerk response to the deeper systemic problem of how to organise and fund a society where public finances are shrinking while demands increase”. Additionally, he points out that a taxation on robots is not directly comparable with humans due to the “burden fall[ing] on the company that has acquired the robot rather than the employee”.

Case studies

Telefonica O2

According to a 2015 case study by professors Leslie Willcocks and Mary Lacity (who specialise in RPA), the UK arm of Spanish telecoms operator Telefonica achieved significant cost savings on its back-office functions by using RPA to automate the swapping of SIM cards and refunding pre-calculated credit to accounts. The project first deployed 20 robots but after initial success this was increased to 160, with the robots performing over 500,000 transactions per month and being supported by just three full-time staff.


Sanoma, a media company incorporating newspapers, magazines, TV and radio, engaged with RPA vendor Blackline to automate a number of tasks related to the manual extraction of data for reconciliation purposes. This was a particularly time-consuming exercise, which needed to access associated data from Excel and SAP, alongside several applications.

Incorporating RPA to fulfil these tasks significantly reduced the completion time and increased accuracy. Otto Sillanpåå, Sanoma’s finance development manager, said of the project: “We reconcile more accounts than we used to, but we use 83% less time. We were able to take out all the manual copy and paste tasks.”

The RPA project resulted in 86% of accounting reconciliations being automated, alongside 69% of SAP close activities.

Sillanpåå’s recommendation to other companies considering RPA was to explore their options sooner rather than waiting for a perfect moment to start.

“If you do it now, you can do it on your own terms. If you wait three years, then you’ll suddenly realise that the race has started. A lot of people are waiting because they want to get it 100% right. Our view was to get 85% right and then 15% can be fixed on the go,” said Sillanpåå.


RPA is also being adopted in the not-for-profit sector, where organisations are often measured on their ability to demonstrate value for money.

Woldmarsh, a not-for-profit farming co-operative, worked with RPA vendor V1 to automate the scanning and processing of invoices. This was introduced to cope with rising demand, with invoices increasing from 11,000 to 19,000 over a three-year period to 2017.

This put the company under pressure to ensure that they were continuing to provide their farmer members with savings by purchasing agricultural inputs such as fertiliser, seed and fuel in bulk.

V1’s software scans each invoice, capturing the data using optical character recognition (OCR) and then uploads it onto a website accessible by Woldmarsh’s members.

This has enabled the co-operative to double its volumes of orders without increasing headcount, as well as letting customers immediately know the cost and specification of their orders.

“Not only has it given us greater visibility around authorisation, it has sped up processing,” said Hazel Copeland, Chief Financial Officer at Woldmarsh.

This resulted in tasks which previously took a week to complete being reduced to a couple of days.

Automation in finance functions

A lot of successful automation is incremental and highly targeted, according to a collaborative thought leadership report published by ICAEW and the Institute of Chartered Accountants of India (ICAI) in September 2020. The report − Automation in finance functions: lessons from India and the UK − includes case studies and practical insights from finance functions using automation technologies, for example, to deliver process efficiencies and applying it to areas such as compliance. 

Useful resources