Start-up of the Week: Froomle improves 'recommendation engine'
A couple of weeks ago, we had the opportunity to feature in the start-up section of Data News digital newspaper. In this interview with our CEO Koen Verstrepen, you will have a better approach to Froomle’s story and our solutions.
Making personalized recommendations is extremely hard. Whether it concerns digital services or physical products, you have to be able to assess your customer. Most recommendation engines are not good at this. Antwerp start-up Froomle tackled the problem using a broader approach.
They have an academic background in recommendation engines and grudgingly saw that things were not going how they wanted them to go. Professor Bart Goethals and doctoral student Koen Verstrepen decided to take a different approach. Streaming services such as Netflix, Amazon, and Spotify are good at this. But if you look beyond those big three, you see how automatically composed lists of the genre 'Recommended for you' often completely miss the point', says Verstrepen.
'You also often see that the personalized offer takes up only ten percent of the screen and is pretty much hidden from view. As a result, in some cases, it only gets one percent of the attention.'
Customized menus
The academics developed a new approach with Froomle in which many more parts of the screen are adapted to the specific preferences of every website visitor. Think of customized menus, banners that immediately draw attention to the right actions and the like. But it's about more than just those recommendations. Based on data, each element can be customized to the visitor's requirements.
These data, for a good understanding, are simply the data that previous visits to the website in question yielded. 'It gives the best results in terms of predictions,' explains Verstrepen: 'Which articles did that visitor read in the past, what products did he or she buy? This means everything we do is also in line with privacy rules.'
A collection of AI algorithms that gets to work with all these data produces various APIs that a customer can integrate into his or her website or web shop. 'Our technology is already incorporated in some e-commerce platforms. All you have to do is activate the functionality and subscribe to our service. You pay an annual subscription fee for this.
From day one, Bart Goethals and Koen Verstrepen were able to sell their solutions and attract customers. 'That was often in the context of a test project', they say. Now our customers include big names such as Mediahuis, Zimmo and Colruyt, and with La Repubblica even an Italian newspaper. The intention is now to quickly scale up internationally because it is our ambition to find entrances worldwide.'
Free trial period
To achieve this, the start-up focuses mainly on its use cases. 'Demonstrating why your solution works so well is the best advertisement', says Verstrepen. This is why we bring stories about our customers, supplemented with information about AI and our activities via our YouTube channel or LinkedIn. To give people an opportunity to test the product, we also recently started offering a free trial period.
Froomle managed to convince a number of business angels from the start, and thanks to them, received a capital injection of a million euros in total. Now that we want to grow quickly internationally, we are launching a new capital round for which we are looking toward venture capital '. The Antwerp start-up is hoping to raise two and a half million.
Note: The blog post is the English translation of the Data News article, written by Matthieu Van Steenkiste.