The display of bounce commerce depends on the solutions used. With AI-based product recommendations, the ad is displayed as soon as a user wants to leave the advertiser’s online store via the back button in the browser. In this case, a redirection is triggered and the user-specific bounce commerce page with the advertiser’s AI product recommendations is displayed. This function is intended to incentivize users to return to the store and make a purchase based on the product recommendations. With other solutions, product recommendations can be displayed directly on the product page, for example. You can find more information in our Overview of the various solutions.

This pop-up – called an exit-intent pop-up – is designed to prevent the user from leaving the page. Bounce management only takes effect when you actually leave the page. So you reach the visitors who are already lost! There is another very important difference: bounce management also works on cell phones. Using the exit-intent pop-up alone means that more than half of the potential addresses are lost. With bounce management you win all speeches.

Our KI works completely without personal or traceable data. Our solution therefore represents a clear contrast to conventional product suggestion systems and KIs, which make their recommendations based on individual preferences. These systems try to recognize personal preferences and then suggest suitable products. We, on the other hand, take a different approach. Our product recommendations are based on the relationships between different products and are optimized by analyzing completely anonymous data such as click-through rates. We don’t track who clicks, only what is clicked. In this way, we can deliver results that are comparable to those of other systems that use personal data – and completely without personal data.

To determine similar products on the basis of artificial intelligence, we vectorize all relevant content using a large language model. This allows content to be assigned to several thousand topics at the same time and placed in relation to each other. An evening dress and a summer dress are thematically closer to each other than a dress and a gas grill. Our system is extremely flexible. We support several established Large Language Models (LLMs) such as Llama2, Mistral, Gemma, Notux and others. These models have different strengths, for example in certain topics or languages, so that we can select the optimum model for each customer. Based on this, we can use any fields from a customer’s product data feed. The results based on product names can be very different from those based on product descriptions. This allows us to find the best results quickly and easily without lengthy training. Finally, all KI suggestions can be further modified. Additional rules can exclude or include products from the result. The AI can also learn from the performance achieved and further automatically optimize the result with regard to various KPIs.

According to a DPD study, on average 62% of customers leave the shopping process. On average, we are able to guide 30% of visitors who leave back to the store and 10% of customers make a repeat purchase. The success of the follow-up page is of course very individual. However, we are happy to offer a trial month so that you can measure and test the effectiveness yourself! We have also received information from individual customers that in some cases over 50% of the customers referred are new customers, which is a high figure in affiliate marketing. When using voucher codes on the follow-up page, we have found that conversions can be increased by a further 30% with a discount code. You can find more success cases in our case studies.