In 2009, Netflix hosted an open competition in search of a collaborative filtering algorithm to predict user ratings for films. Dubbed the Netflix Prize, the contest offered up a million-dollar-reward for the first team able to achieve a 10% improvement of the platform’s recommendation system.
Brazilian entrepreneurs João Bernartt and João Bosco joined the race and with their solution managed a 6% improvement, coming out ahead of all of the other Latin American teams participating. Though they didn’t take home first prize among all participants, the experience inspired them to continue ahead. Bernartt and Bosco created a business plan, and Chaordic was born.
Chaordic transforms artificial intelligence, data mining and machine learning research into high-efficiency recommendations. Focused largely on personalization services specific to the e-commerce sector, Chaordic’s recommendation platform helps clients to sell more, improve conversion rates and communicate better with consumers overall.
Saraiva, Brazil’s largest book retailer, was Chaordic’s first client. The company has since moved on to work for eight of the top 15 e-commerce companies in Brazil, including Ponto Frio, Extra, Casas Bahia and Walmart, as well as numerous smaller businesses in other industries.
Not only does Chaordic provide solutions for e-commerce clients (largely at checkout), but it works with targeted e-mail marketing as well. Its main competitors in the Brazilian market are eBehavior and Tuilux.
João Bernartt, who now acts as the CEO of Chaordic, told us more about the company and provided valuable insights into the e-commerce market as well as recommendation and personalization trends.
Emily Stewart: Brazil’s e-commerce market is booming. However, Chaordic is addressing issues that prevail despite advances. Why go after this specific niche? What makes your solution relevant?
João Bernartt: Big data and context. In the online world, it’s possible to capture, process and deliver a huge amount of information in real-time; however, it demands a lot of investment and focus. This is only possible with big data, and Chaordic’s technology is based on this concept.
That being said, every market has its own characteristics. Brazilians have different behavior when they navigate and buy than North Americans and Europeans. Chaordic was founded in Brazil and knows the Brazilian shopper very well.
ES: You recently participated in a discussion panel on personalization at checkout at an E-Commerce Brasil conference on search and sales. What are the biggest obstacles there?
JB: The checkout area has always been seen as the untouchable page in e-commerce websites. Keeping it simple and fast to conclude the order is essential. However, if you notice that a user has a propensity for buying more, why not offer something tailored just for him or her?
Understanding context is very important. Otherwise, an offer will be seen as a banner or spam, and that’s bad for conversion. The Chaordic recommendation system was developed and has always been improved to comprehend context and offer something only when it is extremely relevant for the user. The results – two-digit conversion rates – have shown that we are doing it right.
ES: There are many startups right now dealing with personalization – recommendations for food and entertainment, content, etc. What tactics and technologies are on trend, and what trends are coming?
JB: Basically, there are two methods for recommendation systems: one based on content, and another based on collaborative filtering. But there are various techniques related to the kind of information that you use to input the data, how you process and how you generate and communicate recommendations. You can apply these two techniques to make recommendations for food, TV, shoes, etc. The key factor for success is finding the algorithm that delivers the best results. Some ideas are excellent and innovative, but if they don’t deliver results, they won’t survive.
The next challenge is to create a fully personalized experience for users, considering all channels (website, e-mail, mobile, call center). You can’t treat your customer differently in distinct points of contact, and, more importantly, you have to know that the customer doesn’t care about channels. Consumers want to have a unique personalized experience with the entire company or brand.