Big Data, Machine Learning and Analytics in the Audit Process

By Bharath Rao, Associate, Quadrisk and Anand P Jangid, Partner, Quadrisk


We are now witnessing a changing trend of conducting business as well as reporting them. This trend has led to accountability issues that have paved way for a high probability of Fraud Risk. Stakeholders are finding it difficult to place reliance on the Financial Statements as well as the processes that are currently followed by the entity.

This has resulted in a requirement of providing a higher level of assurance. We are now witnessing a shift from the traditional test check audit to a 100% Audit. This has also resulted in completely removing the limitation of audit as well.

Higher Level of Assurance is required primarily for the Financial Statements and the Processes and Internal Controls that are present in the entity. A good number of Audit Executives of Enterprises reveal that there have been newer risks on the rise and are finding it difficult to obtain a Birds Eye view of the enterprise on the whole. Challenges such as completeness, correctness and reliability of the data are on the rise. The key challenge here is to correctly consolidate, summarize and analyze huge quantity of data in a systematic manner and to interpret the same effectively.

It is now time for the internal audit function to capitalize on technology as a means to create assurance that gives a 360 Degree view regarding the internal affairs of the company.

The Big Data platform as also branched into this area of assurance as well. Big Data is no longer used by companies to exclusively study the consumer behavior and market trends. This platform has opened up to wider areas such as Risk Management, Audit Risk Management and Fraud Risk Management. Big Data is now being used by enterprises of various industries such as Insurance, Banking, FMCG, Telecommunication, Aerospace, Automobile, Oil and Petroleum, Manufacturing, Steel, Real Estate etc. for Assurance.

By leveraging big data the Audit function will be able to add value in terms of proactive readiness for newer risks, identify gaps quickly and effectively investigate the same. This also results in the Internal audit function to have a greater turnaround time. Platforms and languages such as Qura, Hadoop, SAP HANA, R, Python, SAS, ACL, Idea and others can be effectively implement.

Some suggested use cases of Analytics are –

  • Identification of Vendor Collusion
  • Predictive Analytics for determining the chances of a bad debt
  • Process Mining and identification of process weakness
  • Compliance Management
  • Automation of Internal Controls and it’s enforcement
  • Travel and Expense Claims frauds
  • Governance, Risk and Compliance
  • Identification of gaps and weakness in Material Management
  • Vendor Validation
  • Identification of anomalies
  • Determination of effective point of Revenue Recognition
  • Expense Analytics and determination of provisioning
  • Data Mining
  • Identification of Fraud for promotional items
  • Performance Evaluation against budgeted funds and time
  • Three way match and Payment Analytics

Third Party Public APIs are also leveraged in determining the following pertaining to a transaction –

  1. Determination of transactions with malicious intent
  2. Measurement of value
  3. Duplication
  4. Cross Validation and external confirmation

Using statistical theories play a great role in developing analytical models that would be used to analyze the data. Statistical theories and Methods such as Logistic Regression, Linear Regression, Skewness, Kurtosis, Testing of Hypothesis, Correlation and Statistical Dispersions can be leveraged for developing scripts for data analytics.

Audit data analytics methods can be used in audit planning and in procedures to identify and assess risk by analyzing data to identify patterns, correlations, and fluctuations from models. The analytics derived would be termed as effective only when the data is modelled correctly. Modelling the data is the key to effective Big Data Analytics. Automation is possible once the data is modelled effectively.

Hence it is crucial for the Audit Function to develop effective models that take into consideration all the data, analyze them, obtain better audit evidence and come with better audit opinions on the Financial Statements and Processes and the Internal Controls of the Enterprise.