Tech-driven audit: Knowledge Graphs and the Audit CDM

Franki Hackett - March 12th, 2021

Audit quality powered by the Audit Common Data Model and Knowledge Graphs

This is the second of a three-part series by Engine B’s Head of Audit and Ethics, Franki Hackett, examining the different ways in which technology is improving audit quality. In her latest article, Franki explores how Common Data Models and Knowledge Graphs offer auditors a unique opportunity to focus on what they do best: audit.

The current model isn’t working

The way we currently perform audit is broken. Repeated high-profile audit failures, multiple reviews into the state of audit qualitythe fact that shareholders are less likely than ever to rely directly  on the audited financial statements to make investment decisions shows us that audit is failing to deliver the reliable trust needed.

Sir Donald Brydon recommended we adopt a new understanding of the purpose of audit: “to help establish and maintain deserved confidence in a company, in its directors and in the information for which they have responsibility to report, including the financial statements.”

But auditors face an uphill battle in delivering this deserved confidence. In a medium-sized global company, it wouldn’t be unreasonable to expect more than 150 million General Ledger transactions per year. Even an SME with fewer than 100,000 transactions per year will likely be running multiple systems to store data relating to sales, purchases, payroll, and assets. Where mergers have taken place, some entities will have several different Enterprise Resource Planning (ERP) systems, meaning multiple charts of account and multiple ledgers. With all these systems, even pulling together the basic information needed for audit can take weeks, if not months, of work. From access to completeness and reconciliation, more audit time is spent on preparing data for audit than any other area of work.

When talking about data, each audit is effectively bespoke. It comes in a variety of formats, which means work on data is manual and labour intensive. This causes two problems: a negative impact on audit quality and barriers to competition.

Audit quality is floundering

Audit firms can only spend so many hours per engagement. But delivering a quality audit relies on people: it needs audit juniors to perform good quality testing and audit partners and directors to really scrutinise judgemental areas, challenge management, and consider evidence. This is all takes time. When auditors are having to spend expensive hours on accessing, cleaning, preparing, and reconciling basic data it means they aren’t spending time testing that evidence or making judgements about it. The time required to do a quality audit gets squeezed into whatever is left in the budget after data extraction and tidying is done.

With such massive volumes of evidence now needing to be examined for audit, this is squeezing more and more work into less and less time. As a result, audit quality is floundering.

A lack of competition is stifling audit quality

Because companies now generate so much data, the competitive advantage already enjoyed by the Big Four is further entrenched. Smaller audit firms often do not have the capacity to onboard the level of evidence needed to audit the largest multinationals. Only the Big Four can afford to invest as heavily as they are doing in building analytics from the ground up. But because they (quite reasonably) are focused on developing their own solutions, every time a company wants to their switch auditor, they have to go through a laborious process of getting set up on the new auditor’s systems.

The result? Every firm in the market is either barred from competing or faces enormous barriers to bringing in new clients. Costing more time, more money, and further depressing the ability to deliver the audit quality and reliability stakeholders and shareholders need.

The solution

Audit quality

This is where Common Data Models and Knowledge Graphs come in. Engine B has worked with audit firms of all sizes and industry bodies including the AICPA and ICAEW to develop our open-source Common Data Model (CDM). And we’ve made it available to the market, for free.

A Common Data Model aligns the way we understand and use data, so everyone from the biggest firm down to a sole audit practitioner structures data in the same way. All you need is the mapping that tells you how the client’s ERP system matches up to the CDM. Engine B has maps for the most common ERP systems and we’re currently developing them for the rest of the market. We are also working alongside major ERP developers to make sure ERP systems of the future are built with the Common Data Model baked in.

A Knowledge Graph goes even further capturing the relationships between data points or pieces of evidence. This allows for automatic reconciliations, making sure that data is complete and high quality, and taking the first steps towards automated testing. A CDM and Knowledge Graph aren’t analytics, but they enable easy deployment of analytics at scale, without worrying about data extraction or preparation.

What does this mean for audit quality?

It means no more time spent extracting and tidying data. And it means we can go further, with benefits to audit which are immediate and far-reaching: for example automated general ledger. Or automated three-way matching of transaction to invoice and goods receipt note or work performed, four-way matching to include sales or purchase orders, five-way matching to include contracts or even six-way matching to include confirmation back to bank transactions. Because Engine B can read and extract data from invoices and contracts and match it to transactions using Knowledge Graphs, sample tests of detail can be replaced with automatically checking 100% of ordinary business transactions.

And once you have data in a tidy, regular model, advanced statistical or even Machine Learning techniques can be applied immediately and at scale. These tools can highlight areas of concern, and even benchmark against industry peers to highlight fraud risk. Forecasts can be predicted, and risk assessed to give more confidence about going concern, provisions, or asset valuations. All of these options are available to any auditor with even the smallest firms able to onboard and analyse the largest datasets.

Audit quality

Instead of being limited by how many junior staff are available to comb through data, firms will be limited instead by the quality they offer, and their expertise. Through the use of Common Data Models and Knowledge Graphs our expensive, talented, and highly trained expert staff can be freed up to do what they do best: audit.

Learn more about the future of tech-driven audit by joining Engine B on Tuesday 30th March for our live digital event with guest ICAEW speaker, Robert Hodgkinson.

Share this...
Share on LinkedIn
Linkedin
Tweet about this on Twitter
Twitter