At Eyrene, while working with different clients we face issues of fraud in one way or another. In this series of posts, we'll share our opinion based on our expertise, discuss fraud and anti-fraud, and reveal measures that can be both technical and organizational at the company level to reduce the occurrence of fraud.
We are going to discuss the topic of fraud in relation to digital merchandising systems similar to Eyrene, which are able to recognize photos from a store, analyze the quality of goods display, and how the goods are presented if the brand standards are followed, and finally calculate the KPIs.
We are planning a series of posts about fraud. Post 1 is about fraud and why it occurs. Post 2 will tell about types of fraud and how to prevent it. Post 3 will focus on the following aspects: whether we can beat fraud, how corporate culture affects the level of fraud, and the profiles of Eyrene customers who have a low level of fraud.
Let’s define what fraud is. In short, fraud involves manipulating photos to deliver better results than those that exist in reality. Fraud is an attempt to cheat the merchandising system in order to influence the result. Usually, the goal is to improve someone’s performance, primarily because there is a financial reward associated with it. Fraud occurs when a person cheats the system to make more money.
And what is the level of fraud? It is a percentage of visits in which any type of fraud is detected. Let’s suppose we have 100,000 visits per month. If we get 1,000 occurrences of fraud, then the level of fraud is 1%. At Eyrene, we deal with projects that have different levels of fraud; some may have low levels, some high. But anyway, we ALWAYS get fraud.
Why do people cheat? There are projects where the task is simply to audit. An auditor comes to the store, takes photos, and is not responsible for the result. His task is merely to take good enough photos for auditing. That's it. He just takes photos that reflect the situation in the store. In such projects, the level of fraud is low. There is no need to cheat.
But in most cases, digital merchandising systems are used for quality control. The person who uses the app is directly responsible for making sure that everything in the store is as good as possible. This person takes photos and gets real-time feedback about what's good, what's bad, and what the results are. The feedback is formulated as intelligibly as possible so that the person understands whether he is doing well or not. Ideally, it is a single indicator, such as a score. For example, say the maximum is 100 points, and a person gets 76. The points have a direct influence on the payment. In this case, clients who have implemented a digital merchandising system want to pay only for quality work. Not for the time spent, but for the results. And the results are reflected by the photos in most cases. The person sees the score in real-time and understands that it's tied to the payment he gets. This is the most effective scenario for using the solution. But the downside is that it creates motives to cheat in order to increase the scores. The score can be increased either by working or by cheating. This configuration of real-time communication is the most effective. The ROI is maximal here, as we’ve seen in many projects, but there's also maximal fraud.
In the most effective projects, there is a strong motive to cheat, because people get paid for the results they send with the digital merchandising system; they don’t just do auditing. Secondly, if all the monitoring is done automatically without human involvement, then it is possible to cheat because no one is cheated specifically and some people may feel that cheating on a computer or digital system is not really a fraud. Here is how we get fraud and have to deal with it further.
The vast majority of people are conscientious. Automation of monitoring employees with fast feedback tied to the money they earn generates fraud. This is the main idea of the first post about fraud.
Next, we'll talk about the types of fraud and how to deal with them.
