REAL ESTATE
Driving Marketing Performance with Machine Learning
95%+
lead scoring accuracy
500K
cost savings annually
80%
80% reduction in faliure engineering and model tuning time
Real estate marketer improves sales lead quality using H2O.ai and AWS - Machine learning brings new efficiencies, better insights, and increased revenue
G5, Inc. is a leading marketing optimization company for the real estate industry. Through its Intelligent Marketing Cloud, G5 helps customers optimize advertising and lead management to increase marketing efficiency and effectiveness.
G5 works with more than 7,000 properties in the United States and Canada. Its customers are leasing companies for large apartments, senior living, and selfstorage complexes. G5 employs leasing agents who follow up on leads through phone calls. Unfortunately, according to industry research from Conversica, only 38% of real estate companies can follow up on all their leads, and G5 found that just 14% of leads—1 in 7—were productive. This low success rate resulted in low job satisfaction, high turnover for leasing agents, and low conversion numbers.
G5 wanted to solve this by using machine learning; the company knew that machine learning could help identify stronger leads that would more likely result in sales. Although G5’s product team consisted of trained statisticians and behavioral scientists, the company didn’t have dedicated data science resources to create the needed machine learning models.
H2O Driverless AI allows us to not only take the forms that our customers fill out, but also the phone calls. Phone calls are really important in marketing. Just to give a number here, about 90% of the leads we generate are call based leads. With intelligent core scoring powered by H2O Driverless AI and using H2O Word2Vec models, we came up with a powerful solution that made it possible to achieve our goal of accurately identifying high quality leads from phone calls for our customers faster than ever before."
Martin Stein, Chief Product Officer, G5
G5 identified higher-quality call leads using H2O.ai and AWS
G5 found that H2O Driverless AI on Amazon Web Services (AWS) addressed its challenges with identifying the difference between a productive lead and a dead end.
“At G5 we are leveraging AI to guide the decision-making process in real estate marketing with the help of our Intelligent Marketing Cloud platform that maximizes marketing effectiveness and efficiency,” said Martin Stein, Chief Product Officer at G5. “The G5 Intelligent Marketing Cloud continuously and efficiently improves its accuracy and predictive qualities.”
G5’s first task was to build data sets consisting of 100,000 lead call transcripts and their scores. The company stored these data sets on Amazon S3, and powered its machine learning with the compute capacities of Amazon EC2. G5 then used H2O Word2Vec to analyze the data sets and generate a table of features to serve as the underpinnings of the emerging machine learning model.
![G5 metrix](/ko/case-studies/driving-marketing-performance-with-machine-learning/_jcr_content/root/container/section_460180303_co/par/advancedcolumncontro_1553758322/columns0/image.coreimg.png/1649103949054/g5-metrix.png)
![G5 metrix G5 metrix](/ko/case-studies/driving-marketing-performance-with-machine-learning/_jcr_content/root/container/section_460180303_co/par/advancedcolumncontro_1553758322/columns0/image.coreimg.png/1649103949054/g5-metrix.png)
Having a preliminary matrix of the model, next G5 used H2O Driverless AI to further engineer the model’s features, and train it using the existing data sets. As a result, the model identified high-quality leads with increasing accuracy.
Lastly, G5 needed to make its results production-ready and usable by leasing agents. To do so, the company ran the modelling results on AWS Lambda and passed them through H2O Driverless AI’s automatic scoring pipelines. These are essentially a variation of MOJO scoring, providing an easy, high-performance and scalable way to deploy and display modelling results. The model scoring and complexity was completely removed from leasing agents’ view, yielding a list of high-value leads for leasing agents to contact.
The benefits for G5 and its customers
Model development Using H2O Driverless AI, the G5 team was able to dramatically decrease model development time by 80% and increase the accuracy of lead scoring for customers by six times over traditional lead follow-up. With the speed of model creation and deployment, the product team estimates they are able to deliver the work of two additional senior technical employees without any dedicated data science or deployment resources.
Model results The G5 team increased the accuracy of lead scoring to over 95%. As a result, leasing agents connect with qualified leads 85% of the time, a substantial improvement from the previous 14% benchmark.
Customer results Leasing agents are better equipped to meet their sales quotas. This has a significant impact on increasing leasing agent job satisfaction and decreasing agent turnover, which leads to dramatic cost savings in the sales process. For leasing companies, having more effective leasing agents who stay on the job longer means that they need fewer agents to meet their goals and they can deploy resources to other areas of their business. One G5 customer saved over $500,000 annually.
![AWS](/ko/case-studies/driving-marketing-performance-with-machine-learning/_jcr_content/root/container/section_460180303_co/par/advancedcolumncontro_1553758322/columns0/image_409980747.coreimg.png/1649103949064/aws-lambda.png)
![AWS AWS](/ko/case-studies/driving-marketing-performance-with-machine-learning/_jcr_content/root/container/section_460180303_co/par/advancedcolumncontro_1553758322/columns0/image_409980747.coreimg.png/1649103949064/aws-lambda.png)
H20 Driverless Al: accelerate your Al strategy
H20 Driverless Al employs the techniques of expert data scientists in an easy-to-use application that helps scale your data science efforts. It empowers data scientists to work on projects faster using automation and accomplish tasks in minutes rather than months.
Features include:
- Automatic feature engineering to extract the most accurate results from algorithms.
- Multiple methodologies for interpreting and explaining the results of its models.
- Automatically generated scoring pipelines. Both Python and ultra-low latency.
- Accurate time series capabilities to optimize for almost any prediction window.
- Automatically generated visualizations and data plots of the most relevant data.
More Customer Stories
![](/ko/case-studies/driving-marketing-performance-with-machine-learning/_jcr_content/root/container/section_copy_copy_co/par/advancedcolumncontro/columns1/simplecard.coreimg.png/1649103949125/cisco-end-card.png)
![](/ko/case-studies/driving-marketing-performance-with-machine-learning/_jcr_content/root/container/section_copy_copy_co/par/advancedcolumncontro/columns2/simplecard_copy.coreimg.png/1649103949133/paypal-end-card.png)
![](/ko/case-studies/driving-marketing-performance-with-machine-learning/_jcr_content/root/container/section_copy_copy_co/par/advancedcolumncontro/columns3/simplecard_copy_1925.coreimg.png/1649103949190/kaiser-permanente-end-card.png)