A few days ago, while cooking dinner, my partner (who is a financial auditor at one of the big firms) started walking me through the rituals of her work.
She explained concepts like materiality, sampling and described the long meetings where teams debate what counts as significant, the careful paperwork that has to be produced, the signatures and checklists, the formality of the process, etc. To her, it was just part of the job. To me, it sounded like an entire universe of rules built around deciding which fragments of a company’s financial data should be examined and which ones to ignore.
I nodded along, curious, but my brain was making its own connections. I happen to be in the middle of a project integrating one of my customer’s Xero accounts into a custom dashboard.
The contrast can’t be bigger. On my screen, the numbers reconciles themselves automatically. Data flows cleanly into charts at the moment it’s entered. If I want a report, I can just pull it in seconds. There is no materiality threshold, no sampling, no “professional judgment”, no waiting. It is just… the entire dataset, clean, live and accessible.
Hey, and then it clicked! Her world is built on the limitations of humans. Mine is reshaping other’s by software that doesn’t have those limitations.
About Sampling and Judgment
Auditing exists because, historically, you couldn’t check every transaction. Companies generated mountains of records on paper, and the only practical way forward was to take a sample, apply expertise, and make assumptions. It is slow, expensive, and subjective. But it is the best we had… Until now.
Professional judgment is an auditor’s secret sauce. You trust auditors because of their experience, their eye for detail, their sense of what is “material” and what is not. But judgment is also fallible. It’s inconsistent. And sampling, by definition, never tells the full story.
About Full Data, All the Time
The dashboard I’m building with Xero isn’t just convenient. This dashboard is a snapshot of what it’s coming. Nothing is lost, nothing left unverified, nothing waiting to be checked months later. Everything is validated and reconciled as the data goes into the system.
So, I can see where this is going. With machine learning and big data tools, we’re moving toward systems that don’t just record numbers, but continuously analyse them and flag anomalies instantly. Those systems also catch patterns a human would never notice. This raises an uncomfortable question: Whose output will people trust more? The auditor making judgment calls on a fraction of the data? Or the machine that processes 100% of the inputs, consistently, in real time?
To me, even now, the answer feels inevitable. Trust is shifting towards what the machine says.
The Audit of Tomorrow
This doesn’t mean auditors will dissapear overnight. There is just a massive structure supporting big consulting firms and people won’t shift the trust instantly. But, yes, the centre of gravity will move.
I believe the real work won’t be in sampling numbers anymore. It’ll be in auditing the systems themselves. Like to my partner, I recommend anyone in the field to move from financial auditing to system auditing, where you would audit the actual process of the financial software rather than the financial documents created manually by accountants. Validating that the software is reliable, secure, and fair. In other words, learn to audit the algorithm instead of the paperwork.
Big data and machine learning are changing the world, and the audit of tomorrow will be about code.