Each week 2 million unique visitors browse real-estate website Hemnet.se to look at images of houses. We decided to target these visitors in a relevant context and inspire them with smart solutions for their homes. With millions of images on the site, and thousands of new images added each and every day, we had to come up with an automated way to identify the content of the images, tag them and then match them up with relevant banner content. We managed to do this through the use of image recognition technology and databases. We have actually managed to create a banner ad campaign that adjusts its message to what the user is looking at.

Through image recognition we have succeeded in creating a banner ad campaign which in real-time can adapt its message to what you are looking at. When entering the the real estate pages on Hemnet.se the banner ad is showing products that seamlessly fit in the environment of the real estate shown in the main page. Whenever you switch image (each piece of real estate is presented with a carrousel of five different pictures), the banner instantly switches message – giving you a real time response in a truly contextual surrounding. In order to do this we used real-time user data from Hemnet.se, API:s and we created a database with a vast number of messages being able to map into the specific tags that the image recognition found in the images. The banner ad we have created can adapt its message to millions of images in real-time.

We have created a banner ad that can adapt its message to millions of images in real-time. Even though the prime objective was to drive brand value and inspire users with the vast number of products Vattenfall is offering, we saw high engagement for the smart home products. 20 % of the people who went from the banner to Vattenfall.se soon came back to read more about the smart home solutions. The use of data enhanced the user experience since users were only shown banners that were truly relevant to them in their current online behaviour.