We attended the ICMA Spring conference in Belgrade earlier this month. More than 100 executives traveled to the Serbian capital for the conference from May 10 to 12. As can be seen here, the program was styled for hands-on learning and sharing of experiences. Here are our observations and main take-aways from the many presentations we listened to and one-on-one chats we engaged in.
The main message of the conference probably was: Data is at the heart of fast-growing companies. Smaller companies might not be able to compete with the tech giants on data, but they can at least find their niches and survive very well.
Davor Anicic, business development and sales manager at Styria Digital Services, talked about the many ways image recognition can enhance the user experience. He explained how the “fashion cam” works in the app of Willhaben.at, Austria’s largest horizontal classifieds site. The user can take pictures of clothing and let the app search for similar items on sale. Or, he can pick an ad and start the search from there. “Visual browsing is more popular than taking a picture and using it as a base,” Anicic explained.
Bojan Leković, founder and CEO of KupujemProdajem.com, Serbia’s biggest horizontal classifieds website, showed that investing in machine learning can even make sense in a market with very low labor costs. His company uses it for categorization, but it could also be deployed to detect misuse. However, this is tricky in Serbia: “We can’t act against misuse before an item is posted. Due to government regulations, we would become reliable for the content of the ad which poses too much risk.”
Anna Polishchuk, co-founder at Lalafo, a global mobile c-to-c marketplace based in Ukraine and active in four countries, explained why VC’s might be acting quite rational when they invest so much money in mobile classifieds start-ups. There are lots of challenges, but it’s also a big opportunity:
+ Only 1.5 percent of the population already sells on classifieds sites;
+ Mobile has no difficulty in turning users into sellers;
+ It’s possible to generate GMV at high speed;
+ It’s a sustainable business: If you manage to achieve 10 percent growth, 80 percent of the listings will be created by returning users.
Polishchuk admitted that you need to have a marketing budget right from the very start. Growth usually expands the market, and doesn’t come at a cost to the existing players (It’s not a zero-sum game – editor). For their own expansion, advertising on TV has proven to be most successful. On average, it takes 18 to 24 months to achieve critical mass, she said.
Technology, and especially machine learning applied to mobile, has the potential to revolutionize the way people buy and sell stuff in the long run, said Polishchuk: “Smart assistants know what you own, can assign categories, suggest how to price it.”
Mats Julian Olsen, data scientist at Schibsted, showed how to use machine learning to help sellers get the pricing for used cars right. “Pricing an item is the single most important variable for a seller in a marketplace”, he said. He demonstrated what they are currently doing at Finn in Norway to suggest the fairest price. The idea: help the user towards realistic expectations, and he’ll reduce friction in the transaction. Schibsted wants to do this on all of its 30+ marketplaces. To do that, it needs to centralize data.
“Car prices are a local problem,” said Olsen. Price can vary greatly. He suggested starting with your own, structured data. In his eyes, local methods are good enough, but companies need to invest in infrastructure. The new tool will be available for free to sellers already now, and could also be of value to buyers.
Lots of open issues stirred discussion at the conference: Do people really want to buy stuff on classifieds sites at exactly the right price? What will differentiate classifieds from e-commerce in the long run, if everything is calculated and there are no more bargains? Can smart assistants actually help you discover stuff you like based on analyzing your preferences?
While enhancing the user experience is always a worthwhile discussion, the opportunities for generating revenue is at least as good.
Stefan Salom, co-founder of Infostud, a Serbian company founded late in 2000 that owns several classifieds verticals, shared his strategy of adding new business niches with business concepts close to existing, successful verticals. So far, the strategy has paid off and gives InfoStud confidence to try its hand at new stuff. In 2016, it invested 35 to 40 percent of its profit in new businesses – one of them was the real estate vertical 4zida.rs. Results so far: Visitor numbers grew 4.5 times y-on-y, revenue grew 100 percent, and breakeven is expected around 2019.
More data and revenue topics will be discussed at the ICMA Autumn conference in Barcelona, Spain from October 18 to 20. Block the date in your diary already now!