17 January 2022
Jerome Smail, business journalist, speaks to a panel of experts about how payroll professionals can highlight their importance to a business and how technology such as artificial intelligence (AI) can be used effectively
I and robotics are increasingly relieving payroll professionals of administrative duties. But just how important is AI and automation in payroll, and how can professionals in the industry work with it? What’s more, how can payrollers increase their value to a business in the face of this rising technology?
To find out, I spoke to some leading voices in the fields of payroll and technology:
Ant Brassington, chief product and technology officer at Moorepay
Pauline Green ACIPP MBCS FMAAT, head of product compliance and programs at Intuit
Stuart Hall MA PGMdip MCIPPdip, CIPP non-executive director
Dawn Jackson ACIPP, payroll consultant with Aurore Services
Jaspal Randhawa ChMCIPPdip, EMEA head of products for OneSource Virtual.
How can companies ensure the ethical treatment of payroll records by an AI system, and what levels of liaison should exist between the company and the AI provider?
Ant Brassington: Correct design processes are key, but even more critical than this is the time investment into training the system algorithms. Of course, using an AI provider who has been well-researched to have the right level of security coverage and general data protection regulation (GDPR) terms is also vital.
High levels of liaison must exist between the company and the AI provider – it’s a partnership. The algorithms must be trained by payroll teams and a provider jointly, because each company’s processes will have subtle differences.
Pauline Green: AI is not a separate system per se, but more part of existing systems whereby AI is a tool that helps software provide information the customer can use. Ultimately, the payroll records are the employers’ data and payroll software is only the means of delivering the information, whether that is to the employees (via payslips) or Her Majesty’s Revenue and Customs (HMRC), via real time information (RTI).
All good AI systems should have ethical practice built into their systems. This should include requirements such as not perpetuating social biases or harming end customers. AI should only utilise atomised data. It can then help identify things such as industry trends, or salary benchmarking, which will help payroll professionals.
If you have any doubts about your supplier and how they are utilising AI, then ask questions. Ask to see their privacy and security policies. Also, do they have data stewardship principles? Ask to see those as well.
Stuart Hall: As payroll professionals take care to protect data, by creating a secure physical environment with limited access, ensuring their choice of software is protected by firewalls and password protection, it becomes even more essential to ensure the ethical treatment of payroll records in an AI environment.
AI products can be ethically developed but unethically deployed. It will be essential to form a strong relationship between company and AI provider.
Jaspal Randhawa: It’s important companies understand how data is being gathered and realise that AI-based output is prone to inaccuracies and bias. With this in mind, ask yourself what is the problem you are trying to solve. For example, how often does a process or task need to be carried out? Also, AI offers consistency, therefore, will this benefit the process or task? To what extent will manual intervention still be required? After all, payroll records contain sensitive data, therefore accuracy and legal compliance is essential. When considering the output of AI-driven data, who is the recipient of this data? Do they need to be educated on how this data has been gathered and the purpose it is intended to serve?
Answering these questions will ensure the appropriate solution has been selected and necessary controls established, allowing companies to create the appropriate policies and training material. Going forward, it is paramount AI data output is reviewed to ensure it’s still meeting its intended purpose, that controls are still adequate and there are no compliance issues, such as a GDPR breach.
Any reputable AI solution provider will have robust policies in place backed up by a compliance team, who will work with engineering and product teams to ensure solutions are designed within a compliant and ethical framework. Once a solution has been released or updated, any supporting release notes should confirm the intended use and include a section on ethical use of data (which should have been signed off by their legal/compliance team). If the AI solution provider also offers a consultancy service for deployment of solutions and/or software updates, then consultants should receive the appropriate training. This best practice knowledge should be shared with customers through training material, which will provide more detail than release notes and would help those companies who are new to AI.
Solution providers should ensure customers have the appropriate means to raise any bugs or issues detected with the AI solution. If any of these bugs highlight a compliance breach or identify an unethical use of data then the solution provider will need to resolve this issue as swiftly as possible. For example, they may need to hard code part of their software to ensure customers cannot manipulate or unintentionally produce prohibited data outputs, although the solution providers should have taken rigorous quality checks prior to releasing software.
Solution providers could help further by offering health checks to customers; this service would ensure the AI solution is still fit for purpose and the value of the product maximised.
We’re living in a world where technology is the driver to providing instant outcomes, and where AI is applied correctly, it will result in a better experience for employees as well as company decision makers.
Good payroll professionals are often developed through learning the operational aspects first. Is there a risk that too much automation can leave knowledge gaps?
AB: For the intelligence to run smoothly, a payroll professional needs to train it. For instance, stopping actions and preventing the system making wrong decisions that lead to costly errors. So, payrollers still need a comprehensive understanding of the operational ‘basics.’ However, thanks to increased automation from AI, they’ll also have more time to carry out the strategic elements of payroll.
PG: This is something I have discussed with fellow payroll professionals many times. Those of us who have lots of knowledge and experience are now decreasing in the workplace. There is not perhaps the same desire to learn the nuts and bolts of things. Research has shown younger generations have a different way of learning to older generations. This does not just apply to payroll, of course.
However, as part of this generational change, the payroll industry does need to maintain knowledge, including how to calculate payroll manually, and how to prioritise calculations.
When I recruit to my team, I always test that they can do manual calculations. My team need to be able to do this so we can then tell engineers how to program the software. If this knowledge is lost, then the quality of software produced will decrease.
Automation should help payroll professionals, but it should not be left to totally replace knowledge and experience of qualified payroll professionals.
SH: There is an argument that suggests all payroll professionals should learn how to calculate a payslip manually in order to understand the fundamentals of the calculations.
I firmly believe this is essential.
I recently took my car to a garage. When pulling away from a standing start sometimes there would be little power and no matter how much I stepped on the ‘gas’, it had no effect. Three mechanics examined the car, but were lost for an explanation – that is, until an elderly fitter walked across and within minutes found the problem and put things right. How did he know? The difference was he had worked on car engines long before they became electronic masterpieces. There was a knowledge gap in those first three mechanics.
Dawn Jackson: Generally, I have found that most integration and automation is not designed to replace payroll professionals’ knowledge, but to be used alongside them and their extensive knowledge, helping to simplify processes, save time and mitigate the need for the re-keying of data from one system to another.
When I first started in payroll, it was a very manual paper-based process, but a payroll product these days is expected to receive data from a variety of places, which include timesheets, rostering, human resources (HR), employees, tax codes and notices from HMRC. It must also process the payrun calculations for:
- tax
- National Insurance (NI)
- attachments
- student loans
- pensions
- auto-enrolment
- parental pay
- benefits.
It also has to output RTI data to HMRC, journals to accounting packages, payments to banks, pension information to a multitude of providers (in several formats), payslips to employees in secure portals and reports to the business owners/clients. That’s a lot of data flowing in and out of a payroll system, and I’m sure no-one would ever want to go back to manually processing.
Good payroll professionals will always require a high level of payroll knowledge as they are the ones having to deal with the employee/client queries. Even though software is there to help with these things and more, it’s also designed to allow for many options, which can vary per business. The payroll professional will always be expected to understand how the payroll should be set up for a particular business. Just selecting the wrong option when setting up a pension, for example, can have a devastating effect, once an error is discovered some months down the line. It can be difficult and time-consuming to unpick.
In respect of automated tasks, such as automated pay runs, these are primarily ideal for the simplest payrolls – those with few employees, single directors or those which predominantly remain consistent each period, but still require the user to follow the same processes as a more varied payroll. A certain amount of set-up knowledge would still be required, but the idea is based around taking these payrolls out of the users’ workload to allow them more time to focus on those more in-depth pay runs.
JR: Payroll software has always been designed to meet operational demands; by doing so this reduces the risks associated with manual input. This hasn’t stopped payroll professionals from developing their skills; instead, it’s allowed them to focus on resolving complex matters that require human intervention.
Therefore, AI can only serve to enhance a payroll professional’s skill set. In fact, AI can be used to deliver instant results that would take a person ages to analyse. This allows a payroll professional to focus on strategic interpretation of the AI results, thus enabling the company to make quicker or better-informed decisions.
Will AI create the need for payroll roles to be more grounded in system support and development rather than transactional processing?
AB: Payroll professionals have been using payroll software for decades. Introducing AI and robotic process automation is merely another element of the software to embrace and enjoy the benefits of. Of course, like all new technology, it brings its challenges, but supporting payroll software is nothing particularly new for the average payroller.
PG: I see AI helping promote the role of payroll professionals and their skills. We have seen this in the accountancy area. Accountants have gone from transactional processing to utilising their knowledge and skills in accountancy, to becoming key business partners. Payroll professionals can make the same transition.
They can also help with system support and development. It’s about embracing the change and increasing the profile of payroll within the business.
SH: When computers took over the calculation of payrolls from manual systems, payroll professionals were thrust into the need to support the software and develop new skills. Adding AI into payroll processing could be seen as just an added extra to the support already given. A full understanding of how a decision was reached in an AI environment will be important so that payroll professionals can provide support to anyone affected by incorrect algorithms or breaches of data.
JR: Payrollers are already used to working with highly sophisticated software solutions, so using an AI solution should be no different, provided they receive the appropriate training. The knowledge and skills they acquire broadens the payroll industry landscape, leading to new career opportunities for payroll professionals.
With AI freeing time, should payroll departments strive to be more involved in the design of fair and transparent reward processes?
AB: In payroll, much like banking, no one ever says thank you for a timely payment (but they’re quick to shout about mistakes). Yet banks are now moving with the times, to the extent that consumers are now shouting about the benefits of new, exciting banking features.
Payroll has got to do the same, think pay-on-demand, faster payments, and integration with digital wallets – this is the not-so-distant future of employee experience. And payroll departments must drive this forward.
PG: Absolutely they should. Rewards not only have an impact on individuals but on the company. It is totally logical that payroll should be at the heart of the process. Also, when there are queries on the rewards, the first port of call is usually payroll. Utilised correctly, a good rewards process can help with employee retention.
Payroll professionals need to show how knowledgeable they are, and this is an excellent example of how they can show this to the business.
SH: In short, yes. Any perceived unfairness of employee rewards can be the root of disgruntled employees. Unfairness not only creates a climate of distrust but can be totally counterproductive. Payroll professionals are seen by most, if not all employees, as the trusted people to go to, especially when there is a decision that affects their pay. So yes, payroll departments should strive to be more involved.
JR: Payroll departments can bring a lot of insight to the reward process. Often at the tail end of this process, payroll can provide insight on whether the process is working efficiently or needs to be improved. Furthermore, rewards programmes are designed for employees who will often turn to the payroll department first, should they have any queries – again, payroll can share this feedback, thus supporting the fairness and transparency of reward programmes.
Featured in the February 2022 issue of Professional in Payroll, Pensions and Reward. Correct at time of publication.
