Predicting Audit Results—the Most Vital and Foolish Thing to Do
It may surprise you, but trying to predict outcomes and circumstances is second nature to us humans. Why do you brush your teeth? Because someone found out that brushing your teeth, in most cases, prevents you from a lot of pain and a dentist bill. You try to predict value when setting up a standard, sending an auditor to a production site, or auditing your company or supply chain.
Predicting the future is as foolish as it is vital. On the one hand, predictions of the future will never be 100 percent correct. There is always—and I mean always—a certain percentage of risk that things will not play out as expected. The good news, on the other hand, is that you don’t have to be 100% percent sure. You just have to be close enough to succeed.
The 4 Steps to Achieving Your Goals
Prediction is the first part of the four-step learning process to achieving your goals:
After you have made a prediction and know the likely outcome of an audit, you (2) need to decide what you want to do and put it into practice. You may, for example, decide to skip audits for certain companies producing a particular commodity in a specific region and let your audit manager know that she shall schedule the audits accordingly. After some time, you (3) gather feedback from recall data, personal feedback, or unannounced audits. Based on that feedback, you (4) measure the accuracy/quality of your prediction and determine how you can improve future predictions. After revising your predictions, the cycle starts over again.
The 2 Levels of Predicting
There are two levels of predicting: unconsciously and consciously. Our unconscious predictions start when we work fast and don’t have to „think“ much. This is mostly for predictions with low impacts, such as checking our emails, creating the list for the audits tomorrow, or writing bills. We switch to predicting consciously when we expect a real impact from our decisions.
You see another correlation here: predictions seem to be a part of decision making. To make decisions, we have to be able to choose. If there is only one option, there is no decision to make. Decisions hide in questions like:
- What, how, and when shall we audit?
- How do our auditors perform?
- What systematic do we use to ensure the integrity of our brand or standard?
These are the kind of questions that are not so easy to answer and need quite some thinking.
Technology Is Taking Over Predictions
To come up with solutions—and with that, I mean predictions to make decisions—we humans have developed various tools over time. Experience, paper-based frameworks like Failure Mode and Effect Analysis (FMEA) and IT-based tools for analysis and visualization as we know it from Business Intelligence (BI) today. In the 1960s, a new movement started that reached its tipping point in the last years: Machine Learning.
The biggest advantage of machine learning technology is that it outperforms humans in analyzing and combining large amounts of variables and data. In the book Machine, Platform, Crowd, McAfee and Brynjolfsson from MIT show that machines are a lot better in predictions than humans. That supports the data-driven or informed-decision-making protagonists who claim that the future of all companies lies in their ability to gather, analyze, and use data to drive their business.
AI-Driven Solutions to Support Your Decision Making
Over the last four years, Intact and Wageningen University developed not just paper-based frameworks for predictions and decision making but also IT tools. Some of these tools are based on prediction algorithms and can predict audit non-conformities based on historical performances of companies, auditors, certification bodies, and other influencing factors. An exemplary use case of such a prediction algorithm is a risk-based audit checklist, where all the checklist items with a high possibility of a non-conformity are listed and used to focus the audit on the critical sections in the standard. Another prediction algorithm analyzes, finds, and predicts extreme values, companies, auditors, and certification bodies that perform out of the „normal“. These tools enable the TIC industry to advance to the next level of audit management: data-driven and risk-based auditing.
Standards and regulations are designed to ensure that things are handled within certain boundaries and by good practices. Food safety standards, for example, dictate how food business operators have to act in their companies to produce safe food. Food law does that as well. But how can you ensure that everything works as intended? You “simply” measure—for example, the auditor. Mechanistically speaking, an auditor is a sensor that needs calibration. Is he or she unbiased, constant in the findings, and selective for the crucial parts? Finding anomalies is vital but was extremely hard to control because of data-complexity. Fortunately, advanced analytics comes to the rescue.
Intact Data & Business Analytics
Intact has developed solutions to detect and manage failures previously obscured by data complexity. Intact Data & Business Analytics allows you to predict failures before they even happen and to get notifications on a simple dashboard to act on them. Think of establishing expert rules that are checked in every audit: if 2.3 in the standard receives a good rate, 4.5.6 cannot be good as well. Audit integrity will play a significant role in the near future—for example detecting incorrect audits and underperforming companies, auditors, and certification bodies. Intact is well prepared, and with Intact Data & Business Analytics you can be too.
If you are interested to learn more, I am going to talk about audit predictions, data-driven decision making, and risk-based audit management at the Intact Summit 2019 in Graz, Austria, on June 26.