Leveraging machine learning to build a model that predicted whether a real estate agent would value a homebuyer lead
As part of the RevGen Innovation Challenge, a RevGen consulting team partnered with RE/MAX to provide a proof-of-concept solution for a notable issue faced by RE/MAX agents and brokers.
This project tackled a common problem: overly wide online sales funnels. Traditionally, online lead generation made it difficult to give proper attention to all potential customers who came through the sales pipeline.
Our team surmised that machine learning could be leveraged to predict which leads were most likely to be assigned to agents and brokers. This concept would ensure that agents and brokers received the highest quality leads first, increasing the probability of conversions into paying customers, while also helping agents cut down on time spent qualifying online leads.
RE/MAX’s website gathered thousands of customer inquiries, but with the amount of data available, it was difficult to determine which leads were high value, leading to agent frustration, ineffectiveness, and low conversion rates.
The data captured about leads was limited to a small number of attributes, meaning relevant information was relatively sparse.
The information about leads that agents and brokers received was inconsistent, which impacted their ability to respond.
Less than 2% of leads converted into paying customers.
Given the sheer amount of data and inconsistency of it, we determined that a machine learning model would be the best way to ensure that no potential leads fell through the cracks.
First, we needed to predict the likelihood that an agent would accept a lead assigned to them. The model would then use AI to optimize lead offer routing so that agents receive more relevant leads, resulting in higher acceptance rates and, ideally, better conversion.
We replaced the current process that used simple assignment logic with a machine learning model that scored the likelihood that a lead would be assigned to an agent or broker.
A score would be calculated using lead attributes such as geography, interests, and property type.
"We partnered with RevGen in a proof-of-concept to tackle two distinct agile innovation challenges. RevGen brought technical expertise and know-how to deliver an innovative process that leveraged advanced technology and tools. RevGen presented a structured and agile approach for experimenting and learning for our future success, taking into account our team’s technical acumen. Overall, this short engagement was a success that we can use as a building block to transform our operations."
Through the testing of this model, we identified four (4) potential use cases for RE/MAX. All of them would improve the current process by simultaneously lightening agent workload and identifying higher-value leads.
By identifying which attributes were most important, the model provided insights that could be used to improve the websites, forms, and other data collection mechanisms to route higher quality leads to agents and brokers. And as improvements are made to the triggers over time, the models could be re-run to measure the change in effectiveness of each of these triggers.
The model would also be able to rank leads based on likelihood of acceptance by an agent or broker, thereby accounting for what the agents themselves considered valuable. Once that ranking occurred, the highest quality leads would be routed before the lower quality ones, saving the agents time and frustration on following up on lower value leads.
By using a larger dataset, RE/MAX could predict which leads that are most likely to convert and use that information to prioritize those leads. Given the high-touch nature of real estate, this approach would improve the current 2% lead conversion rate.
While this model was built to work with customer data, a similar process could be deployed to identify agent or broker attributes. Then, a new model could be built to route leads to those with higher close rates or specific real estate specialties.
The RevGen Data Quality Engine gave our client the visibility and control they needed to ensure high quality reporting
Buried under false positives and time-consuming, costly research processes, our client needed a modern, automated method of Revenue Assurance. RevGen brought our data expertise and married it to business processes to recover millions in revenue.
Our client desperately needed to update their Geospatial-powered app because of the poor customer experience and needed RevGen’s help to build a new, modern UI as well as the data and web frameworks to improve the user experience.
Get the latest updates and Insights from RevGen delivered straight to your inbox.