![How can AI assist customers in finding the right product, Niklas Renner?](/unternehmen/uploads/images/_750x375_crop_center-center_none/Niklas-Renner-otto-Ki-Assistent.png 750w, /unternehmen/uploads/images/_1080x540_crop_center-center_none/Niklas-Renner-otto-Ki-Assistent.png 1080w, /unternehmen/uploads/images/_1440x720_crop_center-center_none/Niklas-Renner-otto-Ki-Assistent.png 1440w, /unternehmen/uploads/images/_1920x960_crop_center-center_none/Niklas-Renner-otto-Ki-Assistent.png 1920w, /unternehmen/uploads/images/_2880x1440_crop_center-center_none/Niklas-Renner-otto-Ki-Assistent.png 2880w)
How can AI assist customers in finding the right product, Niklas Renner?
A new feature, the "AI Assistant," is launched on the OTTO platform. The test run includes 180,000 products. Niklas Renner, involved in the implementation of the AI Assistant, gives his insights..
Hi Niklas, you are part of the team that has been working on and implementing the new AI assistant on the OTTO platform over the past few weeks. How long did you work on it before the feature could go live?
Niklas Renner: The Otto Group's Digital Consulting department worked together for a few days in a small team, to check whether the idea could even be implemented with generative AI. Various areas within the group work on innovations every day that benefit customers and are intended to improve their shopping experience. We at OTTO wanted to test the new AI solution first. In a two-week sprint we then implemented the solution together on the platform. It was a highly collaborative process that allowed for a quick implementation.
What were the biggest challenges for you during the implementation?
We encountered the usual problems with generative AI, such as answers that seem plausible but are factually incorrect. By focusing more on the contents of product reviews, we were able to handle this challenge.
How big was the team that worked on this at OTTO?
From our development team, we worked with two developers, a digital analyst, our UX designer and me as the product manager on the implementation on the product details page. From the Otto Group team, there were product owners, a data engineer and a data scientist.
We are now starting a test phase with 180,000 products – why this number?
In order to provide the AI assistant with a large selection of answers that are as balanced as possible, it needs many product reviews as a basis. That's why we only use the feature for products with at least 50 reviews – across all categories – e.g. sneakers, jeans, carpets, coffee machines, sofas.
At the start of the test phase, we are therefore making the option available for around 180,000 products . In the test, we want to find out whether customers enjoy the solution as much as we do. At the same time, of course, we want to check whether the AI assistant makes it easier for them to search for information on a product and whether it intuitively helps them choose one.
How much does AI fascinate you and do you think customers will appreciate such features?
What fascinates me is that we can use such technologies to make the shopping experience easier for our customers. I think it's great that we can use the AI Assistant to grant easier access to product information and information from product reviews.
Our most-reviewed product counts more than 40,000 reviews – this can make it difficult for customers to find the information relevant to them.
Using the AI assistant, customers can now ask a question and get the information faster than by manually searching through the reviews. Whether and how the feature is used and whether the answers actually help our customers, is something we can now see in this test. So we can gather early insights and see whether we want to offer a feature like this in the long term.