Polish Start-ups About – Big Data
Big data is one of the hottest technology trends itself as well as a driving force behind many other trends such as Internet of Things, analytics, cloud or SaaS. But what would happen if we stripped the term of its buzz and took a closer look at what it is and how it creates value and makes money? Meet Polish big data start-ups – deepsense.io and Betegy
Speaking generally, the term big data refers to handling and interpreting very large amounts of data in order to extract some value from them – information that may help notice past mistakes or predict future opportunities for companies and individuals. According to data from Wikibon, the big data segment of the IT industry will have grown to over $33 billion by the end of 2015, which translates to a most impressive growth pace of 22 percent. As an extremely wide and flexible field with an almost unlimited potential for innovation, it attracts start-up entrepreneurs greatly. Companies such as Palantir, Domo or Clover Health manage to raise VC investments of several dozens million dollars each at valuations that stretch up to $20 billion(!). Large global companies such as Pandora or Hulu use big data (machine learning) to learn about the tastes of their users and serve them content they may be attracted to. Some Polish big start-ups largely involved in big data such as SALESmanago, Growbots or Brand24 are already conquering global markets.
But what is exactly big data and how does it help companies on a day-to-day basis? And how do big data start-ups organize it all in a sleek and scalable business model? We're talking about it to the representatives of two Polish big data start-ups – Tomasz Kułakowski, CEO of deepsense.io, and Alex Kornilov, CEO & Founder of Betegy.
deepsense.io is a young start-up company supported by the CodiLime software house. deepsense.io develops the Seahorse – an innovative platforms that analyzes big data using machine learning solutions. The platform allows you to blend data from various sources (NoSQL, relational databases, Hadoop, websites) and manipulate them through a single dashboard. With this information you can create models to help you make important business decisions such as which type of advertising to choose, what to invest in or what kind of promotion to offer (based on past decisions by customers).
Betegy is a Polish start-up and an online service that offers a big data solution that can easily catch attention of even those people who have no idea about the inner workings of big data solutions. With Betegy you can predict the results of football matches. Betegy develops a self-learning algorithm that constantly improves itself. It is said to get it right over 70 percent of the time. BETEGY covers 24 leagues and popular international tournaments.
Who needs big data?
deepsense.io is exactly that sort of big data start-up whose sophisticated machine learning solution is difficult to grasp for an average person. The fact that deepsense.io “delivers the capabilities to design, train and easily deploy sophisticated machine learning models to production using a web-based code-free interface” matters only when you actually understand the benefits it brings to companies. To do that, deepsense.io's CEO Tomasz Kułakowski provides us with real situations where deepsense.io proved how useful it can be in a broad range of areas.
“We have been approached by one of the leaders in global e-marketing and asked to build a machine-learning model that will predict CTR (Click Through Ratio) based on type of advertisement. The main obstacle with such a challenge is the number of advertisements, placement of ads and the way they look. Based on training the dataset provided to us, we were able to perform a thorough assessment of the problem to find the best model. More specifically, by using online learning, feature engineering, and neural networks, we were able to provide the client with a trained model. Moreover, the application of our model resulted in a significant increase (over five percent) in client’s revenue compared to tools that were used before.”
The platform finds use in other marketing-related tasks, including direct marketing:
“The deepsense.io team was working with one of the biggest Polish banks on building a machine-learning model of the bank’s customers to predict their likelihood of purchasing a credit card. Based on historical data and with close cooperation with the bank, the team was able to build a model that improved potential client designation by almost 10 percent.”
deepsense.io's platform also made it to the IT security field:
“We have created a machine-learning solution that classified malware into different categories. Our solution was based on ensembling multiple models based on different feature sets. Final model was able to detect and classify malware into the right category with 99,9 percent accuracy”.
Or even to predict... soil fertility:
“The Soil Fertility contest on Kaggle helped solve a real-world problem of sustainable crop production for the Africa Soil Information Service. We scored 3rd place in this difficult and unique challenge. This contest shows that big data isn’t only reshaping big business, it can also aid agriculture by predicting soil functional properties, which is crucial in planning sustainable farming intensification and natural resource management. We developed a predictive solution for physical and chemical properties of soil using spectral measurements that supports essential ecosystem management services such as primary productivity, nutrient and water retention, and resistance to soil erosion.”
With Betegy, even though the algorithm is complicated enough, the case is a lot simpler. As Betegy predicts football results, it's dedicated to people who bet on those and its efficiency and usefulness is solely dependent on the accuracy of the predictions.
“We built an automated, self-learning algorithm which predicts outcomes of football games with a high accuracy. System analyses arrays of numbers and applies advanced statistical analysis on a scale, which is not achievable by a human. We take over 50 thousand football data points (that range from transfers, injuries and yellow/red cards, through coach impact and public news information to even weather and players' characteristics) and apply the algorithm based on mathematical models and neural networks. As a result, Betegy generates the most accurate game predictions and helps our clients make wiser decisions,” explains Alex Kornilov.
All of those examples serve to show why many different companies put so much faith in data. As its size and methods of analyzing it develop, they may correct many mistakes and open new opportunities for just about any company.
What makes a good big data start-up?
For clients it's all about the quality of the service. But what about investors? With so many new start-ups developing big data solutions, how are investors supposed to pick those that are actually worth it?
“The team is the most important factor followed by management skills. Both are crucial for success in any start-up. The expertise and know-how, which each member brings on-board are significant human capital assets that can make a big difference. Cognitive diversity (mixture of different opinions, backgrounds, perspectives etc.) is also important to avoid the risk of groupthink traps as well as the dangers of dominant logic, both leading killers of innovation,” says Tomasz Kułakowski.
Betegy's Alex Kornilov believes that the expectations of investors and customers are more often than not aligned and it's not up to Betegy to search for investors.
“We think that working technology and demand on the market makes a good big data start-up. It may be true for any startup, but it is extremely important for big data companies. In our case one can easily asses the value of our data by accuracy of predictions. Clear communication of accuracy is an essential part of the pitch to investors and customers. Ideally, we don't want to seek out a VC or angel investor directly, we would like our market success story to speak for us and assist in luring a major investor in.”
Business model for big data
A good argument that support Korilov's opinion on how closely the sheer effectiveness of the solution is related to the attention it receives from customers and investors is the big data business model. One of the reason for this is the fact that most of big data solutions are dedicated to business customers (B2B), which make their decisions based on numbers and actual gains a lot more often than on emotions, a more typical trait for individual customers.
“As with any B2B-oriented business, profits have a direct link to a successful track record that yields more client engagements and impacts the bottom line. deepsense.io is present in the B2B market with the structure resting on 3 pillars: products – in a freemium-premium model delivered on-premises and SaaS; services – machine learning focused on deployment, consulting, maintenance, development, integration and predictive analysis; trainings – organized by deepsense.io globally to breed future leaders in Data Science and Machine Learning,” says deepsense.io's CEO.
Due to unique nature of its solution, Betegy offers its solution to individual customers on a massive scale, but has a lot of business customers as well.
“We provide our data to B2C and B2B clients. For B2C we use a SaaS model with monthly (€5,99) and yearly (€44,99) subscription. Our B2C sales are highly dependent on the quality of our predictions. The higher accuracy we have, the more new subscriptions we sell and the lower is the churn rate (the number of users that give up on the service). In B2B we provide customized predictions with individual pricing for media, gaming and sports-related companies (API with data, widgets, apps, infographics, etc.). We provide our data to ESPN (US), Sport1 (Germany), Matchbook (UK), SunLoto (China), Ringier Axel Springer (Poland), Favourite (Australia) and others,” says Alex Kornilov.
Polish big data start-ups going global
As manipulating data with algorithms, unlike e.g. SEO, is a field largely independent of spoken languages, it would seem that scaling big data solutions over cloud is a piece of cake. It does, however, entails many different struggles.
“The complexity of this relatively young and growing business have inherent barriers to entry and challenges such as internal/external regulations; from legal restrictions to corporate governance rules, which don't allow for outsourcing or exporting big data across borders (health care and finance in particular).”
For Betegy the struggle is somewhat lighter.
“Since we are dealing with sports, we don’t need to adapt the data processing for each market separately. However, it is obvious, that you need to adjust UX&UI. Each market is different and requires localization (language, communication, visualization). Therefore, entering a new market requires significant financial resources and should be part of a strategic plan for SaaS companies. Betegy was planned as global brand from day one. The majority of our clients are from England, France, Germany, the U.S. Therefore, we plan to enter new market every 6 months and, as the result, become the number one destination for sports predictions worldwide”, explains Alex Kornilov.
With such a surge in the number of innovative big data start-ups around the world, it's only natural that more of such start-ups emerge from Poland as well. The technical knowledge of Polish data scientists and entrepreneurs should translate to even more of such start-ups in the future. As big data may or already is significant to just about any sort of business in the world, they may find many more or less unusual niches to conquer.
Got any questions? Send an e-mail to the author