How to optimise pricing and assortment in e-commerce: Miton invests in DeepScout AI tool

To make the right decisions about assortment and pricing, e-shops need data from competitors. However, it is neither trivial nor cheap to get it in a reliable form due to the vast amount of data. Let alone cleanse it, pair it and make the right decisions based on it. AI startup DeepScout wants to change that. It can retrieve relevant e-commerce data faster, cheaper, and usually of higher quality or quantity than current solutions. This is because it uses state-of-the-art technologies in AI and machine learning.

"We can regularly monitor and analyse the product offerings of many players on the market and recommend specific actions in terms of product range and pricing to our clients," says project founder Jan Mittner. In the past, he has founded and successfully sold projects such as Jízdomat (2016 acquired by BlaBlaCar) or Proudly (2019 acquired by Welcome to the Jungle).

The ambition is to improve the business of any e-shop, regardless of segment, size or location: "The platform is almost entirely automated with AI/ML and optimised for big data. Thanks to this, we can offer excellent prices to customers of all sizes. In addition, DeepScout can handle several million products per customer. And it does this in every step - from sourcing market data, updating and versioning data, matching products, analysing results, to recommending specific actions an e-shop should take to improve its business," says Jan Mittner.

Product pairing is one of the key benefits of the product. It can pair information from different sources without having any unique EAN-type identifier. "The tool uses state-of-the-art AI/ML technology to the maximum extent possible. This allows it to extract data in a structured form, check its quality, pair products, learn from feedback, and so on fully automatedly. For example, when matching products, it uses all available data, including images from galleries, and decides whether they are the same product or not. Similar to a human. It just does it on a scale of millions of products," explains the founder.

Miton entered the project as a founding investor before the launch. It is investing 200k euros with an option for another 400k, and Tomáš Hodboď will handle the product development on behalf of Miton. He founded the fashion comparison site Glami 10 years ago and ran it until 2022. Glami is now a billion-dollar and profitable company; Hodboď also helped grow the crowdfunding platform Donio. 

"Unique and clean datasets will be a key element in the new era of the Internet, which will be shaped by language models. We now see great opportunities in data specialisation, quality and accuracy. When Jan Mittner came to us a year ago, we started working on validating the solution. Building 'SimilarWeb for e-commerce' is the first step," says Tomáš Hodboď. "But we don't want to stop there. The ability to offer actionable insights over big data is a topic that is also relevant in other areas, such as real estate pricing, leads for sales, and data for investors. Even these can be handled in a fully automated way today."

Jan Mittner has been working intensively on the project since the summer of 2023, and currently, his team consists of five people. Although DeepScout has not yet been publicly available, it has its first paying customers from Czech e-commerce. "Over the next few months, we will fine-tune the product based on feedback from the Czech market and then head abroad. We have already gone there. We are already providing data from foreign markets for several clients, and it's just as easy as it is in Czechia. So our ambitions are decidedly global," concludes Jan Mittner.

🌐 DeepScout website

📸 Photos for download

🤖 Miton and AI - what we do in the area everyone is talking about