Skip to main content
CUSTOMER STORY

Helping more people find their next dream home

10%–15%

Reduction in time to insight

20%

Increase in team productivity

50%

Lower TCO compared to cloud data warehouse

INDUSTRY: Real Estate
CLOUD: AWS

Even before “big data,” data had always been important in real estate, encompassing everything from property and homeowner/renter information to data about finances, taxes and trends. As real estate continues to mature digitally with PropTech innovations, the amount of data has become more of a hindrance than a help. Since Housing.com wanted to continue to be the first choice of their consumers and partners in renting, buying, selling or financing a home, the company had to rethink their approach to data and bring it up to par with the rest of the brand’s digital strategy. Because data analysts in real estate are saddled with billions of data points scattered among numerous sources — yet expected to make accurate pricing predictions, anticipate renters’ and buyers’ next steps, and detect fraudulent payments — Housing.com needed a data and AI solution as sophisticated as the Databricks Data Intelligence Platform to stay competitive.

Data silos threatened data unification and accuracy

Housing.com knew that having properly managed, hygienic data was high stakes in their industry and could result in accurate appraisals and rental rates, improved buyer and renter engagement, and better risk reduction. However, as the real estate industry became increasingly digitized — from offering virtual tours to providing ways to apply for a mortgage online — the amount of data and touchpoints expanded exponentially. To make matters worse, certain Housing.com teams were creating isolated data in silos that did not correspond to the data other teams were producing, posing a threat to data accuracy. This hindered the company’s productivity, innovation and scalability.

Pricing accuracy on their website was another large obstacle and they found it difficult to effectively monitor website traffic trends and make accurate pricing predictions (i.e., for sellers, the range of pricing they can sell within, and for buyers, the range of pricing they can negotiate within) and ensure properties are priced according to market value as data volumes scaled. In addition, the real estate company was experiencing problems with demand forecasting, personalization and fraud detection. When it came to demand forecasting, data teams were unable to detect future demand patterns due to the siloed data, which was also preventing the brand’s ability to leverage machine learning (ML) to power their personalization strategy.

This led the business further away from their cloud data warehouse and closer to other options to cost-efficiently unify their data with analytics and ML. As Nikhil Sikka, Sr. Engineering Manager at Housing.com, further explained, “The cost of storage wasn’t the problem — the cost of computing was way more, compared to other services. So, in moving away from our data warehouse, we actually saved 50% in total costs.”

Upgrading to a data lakehouse benefited multiple teams

Due to siloed teams and disparate data slowing down the process of gathering data analytics and implementing critical ML use cases, Housing.com began to evaluate new data solutions and ML platform options. Ultimately, the real estate company decided on the Databricks Data Intelligence Platform — which also paired well with AWS — as an all-in-one solution to accommodate their diverse teams of technical and business users.

Sikka clarified, “Everything is in the same environment, and we do not have to juggle the data from one place to another place, or create the model in one place and deploy it to some other place. That saves a lot of time and reduces complexity.” By implementing a unified platform built on the lakehouse architecture, Housing.com could better understand how data is defined across teams, which would help the business simultaneously foster team collaboration and scale data practices within a single solution.

Housing.com successfully moved from their cloud data warehouse to Databricks, now using Delta Lake as the foundational layer for storing data and tables in the Data Intelligence Platform. Housing.com can then funnel highly reliable and performant data pipelines from Delta Lake to feed their analytics and ML workloads, leveraging MLflow to train and deploy new innovative ML models to production. The forward-thinking brand also planned to use Delta Sharing to democratize their data and insights with partners to grow their collaborative ecosystem. They also used Unity Catalog to form a central point of access across all of Housing.com’s workspaces, helping maintain governance of the Databricks Data Intelligence Platform and its open source integrations. As a final cherry on top, Databricks’ integration with Tableau — a data visualization solution Housing.com had previously implemented — would handle business intelligence and reporting, using the new ML models deployed by MLflow to generate different dashboards that measured everything from sales performance to budgeting and financial planning.

Driving large-scale innovation while reducing operational costs

Even with spending 50% less on the Databricks Data Intelligence Platform, Housing.com is still seeing ample results. In a short amount of time, Databricks has proven itself to be a transformative solution for the real estate company. Now, Housing.com can leverage the full potential of their data to make informed decisions that drive business success. According to Sikka, “It has been much easier on our teams because now, there’s only one definition of the data, and it’s also not getting hidden in different silos.”

Because Housing.com became quickly connected across teams with the Databricks Platform, it makes cross-functional collaboration easier than ever. Not only did the unified platform reduce the complexity of the company’s data practices, but it also increased data speeds for innovation, time to market and the production of ML models. Due to the improvement of their data storage, sharing and activation, Housing.com also decreased the deployment time of pipelines, reports and ML models by 10% to 15%. For example, ML deployment now takes one less week of manual work. Speaking of ML deployment, the ML model that Housing.com created to detect fraudulent credit card transactions lowered such instances by 0.05%, increasing customer satisfaction and protecting the company’s brand image.

Aside from more efficient ML deployments, the real estate business has also observed other benefits in switching to Databricks, such as its cohesive nature, which expedites collaborative workflows by 20%. Since there is no more juggling between systems, teams can spend more time on market analysis, business strategy and customer engagement. In fact, part of Housing.com’s engagement strategy includes improved personalization around the company’s recommendation engine that suggests properties, and with improved data accuracy and business intelligence, the real estate company has already seen a 5.5% increase in the rate from prospect to lead using this tactic. As for future plans, Housing.com will add more ML models — which are currently under development — to their website. They are also developing a chatbot to expand upon their growing fraud detection initiative — and Databricks will be there to support Housing.com however it can.