Steering sales: Otocash.com enhances customer segmentation and targeting with Azure

Otocash.com is a platform where Türkiye's car buyers and trusted dealers meet for the quickest, most hassle-free offers on used cars. Its goal is to offer a seamless online experience, characterized by fair pricing, instant payment, and a user-friendly interface. “Otocash.com strives to lead the market in vehicle valuation, ensuring the most precise evaluations for second-hand vehicles,” notes Sercan Eryaz, Business Development and Project Management Unit Manager at Otocash.com. 

Identifying real buyers

With accurate pricing, Otocash.com attracts not only customers who are genuinely interested in selling their cars but also those who are only looking for pricing information. This created the company’s key challenge: separating the wheat from the chaff. “Much like a shop employee might guess if a customer is likely to make a purchase or what they might buy, we needed a system to make accurate predictions,” notes Eryaz. “We also wanted to better understand and segment our customers with the aim of crafting specialized campaigns for different segments.”

As a platform hosted on Microsoft Azure from day one, Otocash.com didn’t look far for a solution. “We were drawn to Microsoft due to the continuous service, fast local support, and the ease of using machine learning interfaces, even for first-time users,” shares Eryaz. Initially, Otocash.com started extracting data and conducting experiments in-house. “We couldn’t progress as fast as we wanted, so Microsoft introduced us to Nephos. They provided outstanding support in various aspects, including architecture, screen usage, and code structure,” he adds.

The outcome of this collaboration was an artificial intelligence model, developed using Azure Machine Learning. Otocash.com trained the model with its in-house data to segment customers and determine the likelihood of a transaction. “In machine learning projects, the quality of your data is paramount,” notes Eryaz. “Even with a highly organized data set, examining correlations among data points can be time-consuming. Azure Automated Machine Learning streamlined the process, allowing us to quickly choose the appropriate model from hundreds by simply uploading an output table into the system.” 

Better targeting, rising sales

Otocash.com quickly saw improvements in customer targeting, boosting sales. By allocating the marketing budget to a more specific audience, the company recorded a 14 percent rise in vehicle purchases despite a challenging period in the market. “Overall, the project led to a 20-percent increase in customer conversion rates. Additionally, our system that automatically generates sales offers became 30 percent more effective at turning interested individuals into buyers,” shares Eryaz. 

The algorithm quickly became the talk of the company, sparking friendly internal competition. “We wagered guesses against the machine learning model on whether customers would sell their cars,” recalls Eryaz. “Impressively, the model's predictions were right 60 percent of the time, doubling our own success rate of 30 percent. It turned into both an engaging and enlightening experience for the team." 

The project even reduced the workload of the company's customer service team by 94 percent, as the model accurately identified potential sellers, reducing the need for manual verification. 

Today, Otocash.com continues to fine-tune the algorithm, making its platform future-proof. "Given the evolving landscape of Türkiye's automotive sector, where one can awaken to a transformed marketplace overnight, our system must deliver consistently precise forecasts,” concludes Eryaz. “This requires regular maintenance. Looking ahead, our objective is to automate this framework, enabling the system to update itself autonomously."