Künstliche Intelligenz am Point of Sale

AI solutions are finding their way into more and more industries. Retailers are also investing in artificial intelligence (AI), because it allows them to simplify and accelerate their processes. Last year, the German Retail Association (HDE) asked 1,000 retailers about their attitude toward the use of AI –  20% of respondents use AI or plan to use AI, and just under 70% of those surveyed attach a very high level of importance to AI over the next ten years.

The potential for using AI is huge, including in brick-and-mortar retail. In this blog post, you’ll read how to push your success metrics through AI at the point of sale, detect fraudulent actions during self-checkout or self-scanning while minimizing unnecessary rescans, and optimize markdowns by using Markdown Pricing. Finally, the blog will cover how you can strengthen your customer loyalty through smart incentivization.

Avoid losses through fraud detection

Self-scanning and self-checkout can lead to improved service for your customers, as they avoid queues at the checkout and save time. However, both services also provide opportunities for fraud to occur, which can lead to losses for you. AI-based fraud detection can mitigate these risks and protect customers from fraud while minimizing unnecessary rescans.

Based on customers’ current shopping cart transaction data, AI predicts the probability of an irregularity. If the probability is above a certain threshold, the recommendation for a rescan follows.

Unlike a static or configurable set of rules, there is no need for you to constantly maintain and tune the system if you use an AI-based solution. This is because with a static set of rules, you have to configure new information or changes in the system. If differences between several markets and/or regions are added, maintenance becomes confusing and time-consuming, potentially resulting in misconfigurations. This recommends unnecessary rescans that can harm the customer experience and lead to increased staffing.

Another challenge is the aforementioned rescan of the rule set. If the evaluation of the rescans shows that too many fruitless rescans were performed, this means a negative impact on the customer experience. But what consequences does this have for possible changes in the rules and regulations? Usually, it is difficult to understand which individual criteria led to these decisions and by which adjustments the result could be changed.

This is not the case, however, with an AI-based solution. It works on the basis of the existing set of rules and constantly optimizes them. This is done by providing feedback on whether a rescan was “successful” or not. The AI application is constantly learning and uses this knowledge for future decisions.

In addition, you able to implement your own rescan strategy, e.g., including a trust level, via the extension model of the POS system.

Another advantage of the self-learning system is automatic adaptation to new fraud strategies. There is no need for manual fine-tuning or time-consuming maintenance of the system on your part. As a result, the accuracy of the prediction for a fraud case increases. This allows you to reduce the number of rescans to the lowest possible number so superfluous rescans no longer harm customer perception.

Fraud Detection, Betrugserkennung am POS, Künstliche Intelligenz

Minimize your markdowns with Markdown Pricing

Markdowns mean one thing: losses that reduce gross profit. No matter what industry you’re in, products you don’t sell inevitably result in losses. In addition, there are costs for the disposal or recycling of the unsold goods.

This poses a particular challenge for food retailers (LEH): Every year, a “waste mountain” totaling 12 million tons of food is created in Germany. Most of the waste is generated in private households, but the retail sector also generates around 0.5 million tons a year. Supermarkets face the problem of selling fresh produce before it spoils, and these sales depend on many different factors. For example, the day of the week influences people’s shopping behavior, and fruits and vegetables are available in different quantities depending on the season. Ordering according to demand, based on past experience, is an important component,  but in reality, this estimate is often influenced by unpredictable events and requires a high degree of flexibility on your part.

Markdown Pricing based on artificial intelligence forecasts the exact sell-by date for each product. Depending on how much the forecast deviates from the target sell-by date (in this case, the spoilage of the product), intelligent pricing algorithms adjust the price up or down fully automatically. As a result, less food ends up in the trash. AI can take good account of the special challenges of food retailing – read more about this in the article “5 Approaches to Successful Dynamic Pricing in Food Retailing”.

Dynamic Pricing am POS, Abschriftenoptimierung

Strengthen customer loyalty by incentivizing them with slot machines

Retailers struggle to gain customer loyalty at the point of sale. Only a few customers own a classic customer card, e.g., the Deutschlandcard or Payback card. But this form of incentivization via customer card systems brings a decisive disadvantage: they do not create direct customer loyalty between you as a retailer and your customers. Often, there is also an uneasy “cost-benefit” feeling, since the incentivization in this type of customer loyalty program is based on sales instead of earnings.

The use case “smart incentivization at the slot machine” offers an alternative. This is an example of gamification at the point of sale that helps you encourage your customers to come back.

It works like this: The AI calculates the contribution margin of a shopping basket via the receipt number in real time at the checkout. After shopping, your customers have the option of scanning the receipt they received at the slot machine. Depending on the contribution margin, the slot machine generates a winning amount for your customers between 0 and 100 cents. Customers can then decide for themselves whether they would like to receive this amount as a value voucher or would prefer to donate it to one or more institutions of their choice, such as a local association or kindergarten.

Incentivierung am POS, Slotmachine

Artificial intelligence at the point of sale – further use cases for your success

These three use cases, fraud detection, minimizing markdowns and incentivizing via slot machine, are not the only ways to ensure business success. In our articles AI at the POS Part 1 and AI at the POS Part 2, you can learn more about other use cases for stationary retail from the areas of dynamic pricing, personalization and real-time prediction.

Have you seen another exciting use case? Contact me here, write an email to MrPersonalization@prudsys.com or message me on LinkedIn!


Mr Personalization