Preamble. Reasoning. The Start Bit.

Where it starts.

I'm a musician (kidlupin.bandcamp.com), a chartered accountant and a data scientist (a wee bit). For the past six years I've worked in financial audit in Northern Ireland. I've audited lots of different organisations - companies, charities, public sector bodies, partnerships.

I've come to the conclusion that financial audit needs a bit of a shake up. The introduction of the revised ISA 315 has resulted in a fairly major philosophical re-orientation of the audit approach. The new approach is more explicit on the importance of taking a risk-based approach. This, for the auditor, requires a more profound understanding of where risk lies in the audit of your client.

There's a lot more that could be said here, however this is day one and at risk of becoming an old man screaming at clouds.

I have decided that the best way I can improve audit quality in the light of the revised ISA 315 is to create an open source financial audit tool in R.

Why do I think I can do it?

Firstly, I have the subject matter expertise to enable me to design a piece of software that will perform the relevant analysis. Equally important, I know what can go wrong in the audit process (this is mainly in an administrative sense, for example when your client makes radical amendments to their trial balance halfway through the audit) and therefore build in specific processes within the software to address these.

Secondly, I have some experience in data science and scripting in R. I've spent the past year performing analysis on financial audit data and complementary datasets in R. I've a good idea of the issues that arise when data wrangling financial information (particularly general ledgers) and how I would want the analysis to be presented to the auditor. I had the fortune of attending a session with Andy Kirk in 2022 and that really opened my eyes to the graphical opportunities available to me.

Thirdly, there is a fundamental rule in audit about the primacy of auditor judgement. This is why machine learning isn't allowed (yet). It also means that the ability of the auditor to do their job most effectively is influenced by their ability to access information in a timely manner and in an easily understandable format which they then assess based on their own holistic understanding of the entity being audited. I want a tool that will help me do my job to the highest possible standard. I think other auditors might want a tool like that too.

Why R?

R seems a sensible environment for this project. It's pretty easy to read and understand, it churns out markdown files or similar, which are essential for one's audit trail. I can stick a Shiny interface on it when I get that far. As far as I know I'll be able to tie it in with SQL which will become invaluable when I need to implement multi-year analysis of large general ledgers and supporting datasets. Most importantly, R (and specifically the ggplot2 package) gives me access to most of the visualisations that I think I'll need to represent the data in a way that supports the type of risk assessment required under ISA 315.

Why open source? Why a digital financial audit tool?

The world of auditing and accounting is a profoundly conservative one. It has generally failed to keep up with the pace of change in technologies used by their clients. Many hours are wasted by junior auditors on engagements where they are carrying out mindless administrative tasks. Reformatting documents, inputting formula, summing columns, cross-referencing and so on. Tasks which do not improve their judgement and which carry a level of manual activity such that the risk of error is increased. Tasks which add no value to the audit for either auditor or auditee.

I want to eradicate these inefficiencies from external audit and empower auditors by recovering the time to develop a more profound understanding of their clients, therefore better identifying the risks arising in the audit, whilst minimising the chance of error.

The plan is to write everything in R and release it as is. The initial set of scripts will need to be run in a specific order but this will mean the auditor can only use the chunks that they need for their audit. Ultimately I want to create a Shiny interface but that is a long way away at this point. We need to walk first.

Why LUPIN?

I have called the software LUPIN, after myself of course. It's a fun word. There weren't any other (more sensible) options proposed.

Anything else I want to introduce to the conversation?

Yes; the auditing profession has a complex and unresolved relationship with accounting fraud. On one hand, auditors cling to a comment from an 1896 judicial pronouncement that "an auditor is a watchdog; he is not a bloodhound". As I already said, accountants are pretty conservative. On the flip side, numerous auditors in recent years have been chastised for failing to be rigorous enough in their procedures to identify frauds which have ultimately caused the collapse of large businesses. There is a general sense that the auditor should perform sufficient procedures but not go out of their way (which, for the purposes of this conversation, I will deem anything requiring many hours of complex analysis).

However, what if this analysis could be performed almost instantaneously? Then there would surely be less of an excuse not to do it. Or perhaps it would free the auditor up to pursue other, valid, lines of inquiry into areas more susceptible to fraud such as payroll or supplier details.

Just a thought.

Why this blog?

The purpose of this blog is to record the development of the tools. Like all grandiose projects which are intended to change the world, this will obviously take far longer than I plan. But I want a place to post my thinking (and subsequent amendments), attach my schematics / workflow / scripts / whatever. If anyone reads any of this and wants to chat, leave me a comment.

For what it's worth, I love auditing. It's a fascinating job that can make a real difference to society. I just don't think we are doing it very well. Hopefully I can make a contribution to make it better.