Hong Kong property agent Ricacorp is turning its volumes of data into knowledge that will help its 700-strong property agents to enhance the quality of service delivered to customers.
“We wanted to build a new platform for our internal systems,” said Dennis Tam, associate director of Ricacorp Properties. “Our direction is digital transformation and we are looking at cloud computing, AI, big data and mobility to achieve this goal.”
The 35-year-old property firm has recently rolled out its new Rica+ Artificial Intelligence platform that allows its agents to make smarter decisions and smarter recommendations to customers with insights generated from Microsoft Azure machine learning and data analytics.
Ricacorp has deployed its new system using platform-as-a-service provided by Microsoft Azure Public Cloud Services.
“We have accumulated a lot of data over three decades and we want to use our data to empower our employees,” said Tam.
Ricacorp has a huge property base with over 1.7 million flats, including private flats, public rental housing as well as Home Ownership Scheme flats. It has a total service attendance of over 4.8 million people with nearly one million pieces of company information to be handled monthly.
Tam noted that preparing the data to be able to go through machine learning has not been easy.
“We need to consolidate all the internal information such as current price and feedback from customers. We also have to put gather external information from the Land Registry and newspapers, and feed them into the system,” Tam said.
“We also need to restructure all our services again. Using microservices and containers, we needed to deliver new APIs before we build the Rica+ platform,” he said. “
Building new system using PaaS makes deployment easier and faster, according to Cally Chan, general manager, Microsoft Hong Kong.
“We are not only providing computing resources but also all the APIs and framework on top. And for Ricacorp, it released them from the burden in building from ground up these framework and APIs so that they were able to focus on developing the functionalities,” she said. “Time to market is highly improved and if the time comes and they want to add a new feature, the can do so in several days instead of maybe a month.”
It took eight months to build Rica+ and the first version of the new platform was released last October.
Development is still in progress to deliver its full capability with new features being added and existing ones upgraded on a regular basis.
For one, the mobile interface has been enhanced to make it easier for agents to access the Rica+ app anytime, anywhere.
In addition, data security has been enhanced with the use of face recognition and multi-factor authentication technologies. The embedded GPS function conveniently allows agents to check in real-time at any location without the need to take attendance at the branch manually. This increases the agents’ overall productivity.
“In recent years, biometric authentication by mobile phones has become popular,” Tam said. “The advantage is that biometrics cannot be easily replicated, and any lens-equipped phone can become an authentication tool through an API. Thus, pictures of employees can be used as log-in information, saving time and enhancing convenience.”
Putting AI and machine learning in
Three months after the launch of Rica+, Tam said the system delivers output similar to human judgement.
“We need to clean up the data and add more factors to find the correlation and to get more accuracy. The output really depends on how often we put new features in,” he said.
During the initial rollout, Tam and his team concentrated in replicating features and functions from the old system in a bid to fulfill the agents’ expectations.
Ricacorp expects to invest over HK$20 million this year in its digital transformation efforts using AI and public cloud. This amount is on top of the more than HK$20 million already earmarked for last year’s IT services enhancement.
In the coming months, Tam and his team will be working on bringing out the full potential of the platform's AI and machine learning capabilities.
“We don’t want our property agents to solely rely on their own experience when giving recommendations to customers, as they had been doing for the last 20 years,” Tam said.
As Rica+ matures, it would be able to analyze online data – such as property valuation, recent transaction history, listed price offered by sellers – with internal records liked customer preferences to help agents make suitable flat recommendations, and thereby enhance the sales conversion rate.
Furthermore, machine learning can assist agents to improve matching success rate by giving recommendations based on the analysis and learning of previous similar customers. The system can filter useful property information with customers’ interests and preferences through its collaborative filtering feature.
“After data is processed and standardized, it will undergo repeated tests within the machine learning model to obtain more objective and potential flat selection, which agents can then focus on promoting and improve success rate,” Tam said.
With new AI platform, Ricacorp is able to more effectively use digital marketing strategies, with business revenues and profits expected to increase by 20%-30% and 40%-50% respectively.
In the future, Ricacorp plans to give customers access to the new platform, allowing landlords to update information such as a flat’s price directly into the system.
“We hope this will become a single platform for our agents and our customers so that that we can speed up the whole process of digital transformation,” Tam said.