Looking across the audit industry, Stuart Cobbe – principal consultant at The Analytical Accountant – sees stark contrasts between how small to mid-tier firms are working with artificial intelligence (AI) and analytics tools, compared to what is happening among their larger peers. “There’s a division between audits regulated by the Financial Reporting Council (FRC) and those covered by ICAEW,” he says. “And that broadly aligns with the size and complexity of the clients that are being audited.”

As former global head, analytics and industry insights at leading AI audit software provider MindBridge, Cobbe maintains a strong interest in how solutions such as AI, open banking and analytics tools are helping audit to evolve.

“At the larger firms,” he says, “including the mid-market, adoption of such tools is continuing to accelerate. Indeed, in many cases it’s hitting a critical mass, whereby it’s expected. It’s becoming part of best practice to carry out some form of data analysis across journals – often based upon machine learning or AI – together with a data-driven approach to risk assessment and revenue testing.”

Finding needles

KPMG, for example, is home to intuitive, ‘smart audit’ platform Clara, powered by Microsoft Azure. As KPMG chief technology officer Matthew Campbell explains, the firm has already pinned down a number of use cases for AI in audit – one of which is anomaly detection.

“With traditional analytics tools,” he says, “you would typically set a series of rules and look through a large dataset for any exceptions. But with AI, you feed a population of data to the software and it quickly identifies anything unusual – like you’re finding needles in a haystack, without knowing they’re the needles you’re looking for.”

Campbell also cites document reading and judgmental audit as areas where AI is adding value. “Extracting information from and interpreting large volumes of unstructured data is a growth area,” he says. “On the judgemental side, AI is playing a supporting role in work that involves elements of prediction or validation – for example, asset valuation. So, we’re using AI fairly widely – mainly to automate parts of the audit process, or to supplement auditors when they’re applying their own judgement.”

In that context, Campbell frames AI-enabled continuous auditing – the scrutiny of clients’ statements on a rolling, real-time basis – as something the audit industry is working towards, rather than a use case with tangible, present-day impact. “The technology is available,” he says, “and I’m aware it’s being used to a limited extent on some internal audit engagements – so we’re typically seeing some entities we audit using such technology to provide a ‘first line of defence’ type of assurance, rather than an independent, external form. But there’s definitely scope for it to go in that direction.”

Systemic risks

On a more current note, Campbell says innovative data-crunching tools are helping KPMG auditors to conduct root cause analysis for two risk-based purposes: process mining and trend analysis.

Process mining retrospectively maps out all the routes a transaction or business process has followed, providing auditors with an audit trail. It also detects inefficiencies, risks and errors. That information can then be used to highlight potential for failures on similar processes in the future.

Trend analysis harnesses data to identify trends for comparing similar companies, industries and/or specific processes in benchmarking or progression testing. Through that analysis, auditors can identify developments that may not be immediately obvious to the human eye – but which could reveal the root cause of systemic risks.

“For example,” Campbell says, “you may discover that the client company has an overstretched employee who’s approving lots of inaccurate invoices. As their total workload may involve processing millions of invoices, that suggests a significant issue could build up in the long term.”

As the business environment becomes rapidly more digitised, to what extent does Campbell think new technologies will affect how large firms audit pure-play digital companies – those without physical premises, and with fully digital product lines? “This is about understanding the industry in the same way that auditing an agricultural business is different to auditing a manufacturer,” he says. “Conceptually, the process will be similar. But there will be different risk factors and accounting judgements – for example, no premises means less emphasis on fixed assets, and digital products means greater emphasis on intangibles. So it’s important to bring in the right level of expertise, through both recruitment and upskilling.”

In KPMG, upskilling is also helping to boost auditors’ understanding of emerging technologies – enabling them to audit the various new tools at their disposal. “We have an initiative called KPMG Audit University,” Campbell explains. “Every year, we take our entire audit practice through this scheme, to update it on new standards and technologies. That gives our auditors the chance to get really hands-on with these tools, so they can determine how best to audit them and then use them to deliver high-quality audits in the field.”

Quality gains

At mid-tier firm Mazars, audit quality partner Paul Winrow highlights the benefits of tools such as Confirmation and Circit that are automating bank confirmations – marking a departure from the traditional, paper-based process and its attendant inefficiencies. “Streamlining confirmations has been an important outcome,” he says. “The workflow we have now with digital platforms is so much more efficient than the old system of mailing out hard-copy requests, waiting for confirmations and calling the bank to chase them up.

“But at the same time, we shouldn’t underestimate the quality gains. In my view, confirmations are a critical part of audit quality – some of the high-profile scandals that have hit the audit sector in recent years have revolved around the absence of vital confirmations. So, for me, automating the process is such an obvious step to take. Yes, it’s a lot more efficient – but the timeliness and completeness of responses to confirmations is also important in improving quality.”

Winrow is even more excited about the potential for open banking to drive quality. Last year Mazars, a group of other audit firms and Circit co-signed an open letter to regulators, calling for the data available to auditors under current open banking rules to be broadened to maximise the impact on audit quality.

“If we can get open banking right,” he says, “then from a quality perspective, that would be a game changer. Instead of checking transactions and picking a number of samples for testing, you’re just looking for outliers – and the wider the dataset is, the easier they are to spot. That frees auditors to focus more on the judgemental aspects of the audit.”

New breed

Where could these developments take large firms in the next five years?

“We’ll see the skillset changing for the auditor of the future,” Campbell says. “Client companies are getting increasingly data rich. So we need to have tools that will keep on top of that data and deliver high-quality audits. Technology will never completely replace people – but alongside traditional skills, the new breed of auditor will also need to be part data scientist and part technologist.”

And how could these technologies affect the traditional boundaries between small, mid-tier and large firms? Campbell says: “As barriers to use are lowering, you could argue that these tools are available to everyone. But equally, large firms still have the capacity to invest in proprietary technology. So, I think the nature of the distinctions is changing, rather than the distinctions themselves.”

Prominent futurist and consultant Rohit Talwar, CEO of Fast Future, takes a different view. “Given the liability and scale issues around making AI work well,” he says, “we could see the formation of new alliances between tech vendors and audit experts. As such, in the next few years, we could see some of the most tech savvy partners in the big partnerships break away into a franchisee-style model – operating under the main firm brand, but essentially functioning as fiefdoms. While they may buy in certain services from the umbrella firm, a lot will come from other providers, too.”