Chewy

I 'm an engineer/developer at heart. I love learning and building new things. Teams, tools, solutions, automated processes, dashboards, reports, and insights, the list is endless. I have built and maintained business intelligence and analytic functions multiple times during my career, from concept to execution. Whether it's for internal audit, project/ program management, or operational processes, they all need the ability to pull insight from data and turn it into something meaningful that can help drive the direction of the business or point out key risk areas that need to be continuously monitored and mitigated.

There are four types of common data analytics: descriptive, diagnostic, predictive, and prescriptive. These data analytics also represent a maturity model, with descriptive being the easiest to start with and then maturing to the other types of analytics.

Descriptive analytics

Descriptive analytics are the easiest to implement in internal audit. Descriptive analytics tell you what happened in the past. Descriptive analytics are great for quantifying the value of an engagement letter or determining the scope of an audit. Performing descriptive analytics may include combining historical data from multiple data sources to get a full picture of what happened.

What it may look like in practice: A dashboard or report delivered to the audit team showing how many transactions happened last month and their amount. This information is then used in either the engagement letter or leveraged during the scoping process in order to understand scale.

Diagnostic analytics

Diagnostic analytics provides insight into why something happened by slicing and deciding your data into different views and scenarios. This is where data analysts drill down into the data to find dependencies and identify patterns.

  ​The four types of common data analytics: descriptive, diagnostic, predictive, and prescriptive also represent a maturity model, with descriptive being the easiest to start with and then maturing to the other types of analytics 

What it may look like in practice: In fieldwork, a correlation analysis is performed, and results are delivered to the audit team showing what key factors such as department, store or salesperson are correlated to the transaction amount. Any strong correlations are then called out in the final audit report to strengthen the internal audit's findings and conclusion by proving them through the data.

Predictive analytics

Predictive analytics predicts the future based on the past. Regression is the most commonly talked about prediction model for internal audit, but there are also great benefits in decision tree and clustering models. To predict, you need to have solid historical data and previously identified cases and data of what you are looking to predict.

What it may look like in practice: Using correlations discovered previously, a predictive model is built to tell internal audit how many transactions may happen in stores next month/quarter/year. This data is then used to help quantify the risk level if no changes are made.

Prescriptive analytics

The purpose of prescriptive analytics is to determine what to do to mitigate a risk. Gathering data for prescriptive analysis can be difficult as you may need a combination of internal and external descriptive, diagnostic, and predictive data.

What it may look like in practice: Through the understanding of correlated variables, predictive models, and a-b testing, internal audit has determined what levers the business can push and pull to mitigate the risk of transactions. Internal audit uses this data to provide a recommendation to the company.

Practice being agile by figuring out what works and doesn't work for your internal audit department and continuously adjust your approach. I hope you have found this article insightful and are excited to start on your own analytical journey.