Nearly a year after growing its footprint to one million agents and 300,000 active monthly users, Homesnap has reached another milestone — the release of All-New Homesnap Pro, a revamped, artificial intelligence-powered version of its marketing and productivity app for agents.
All-New Homesnap Pro, which is available beginning Monday to members of more than 200 Homesnap MLS partners, includes a machine-learning algorithm that identifies potential sellers, real-time MLS data integrations, advanced property search features and improved building pages.
“We listened to our agents and discovered that the most requested features for a real estate app are tools that give agents comprehensive property and homeowner information and ways to identify potential sellers,” Homesnap CEO John Mazur said in a prepared statement.
In a phone call with Inman, Mazur said the three-year research process behind creating All-New Homesnap Pro, which included conducting one-on-one agent, broker, and MLS interviews, hosting think tanks, and multiple beta tests.
“One of the areas where we feel blessed and fortunate is that agents will tell you when you’re doing things right, and when you’re doing things wrong,” he said. “We called agents in and had think tank groups, and we talked to brokers and MLSs, and found out what agents thought was missing from the platform and what their wish list was.”
Armed with the feedback from agents, brokers, and MLSs, the Homesnap team became laser-focused on diversifying its data sources and using machine-learning technology to improve the quality and usefulness of the intelligence it offers to agents.
All-New Homesnap Pro’s features are powered by more than 500 data sources that include MLS data, third-party MLS data integrations and public records information. MLS data and third-party integrations from CoreLogic and Black Knight support the app’s new, real-time MLS data integration that syncs contacts, and saves searches and listing carts.
Meanwhile, MLS data and public records are responsible for the app’s advanced search filters that include criteria such as parking needs and specific appliances, and building searches that allow agents to view and compare multiple units at once.
“We want them in one app to be able to pull up a 360-degree view of any property, on market or off market, and be able to take action with intelligence that’s actionable,” Mazur told Inman. “Agents spend the most time in their MLS platform and Homesnap is second. So we want it to be this seamless, integrated experience.”
Although Mazur and Homesnap Chief Product Officer Lou Mintzer are proud of each of All-New Homesnap Pro’s features, the app’s Off-Market Marketplace is the crown jewel. Mintzer described it as an advanced adaptation of machine learning.
“We built a feature that uses a predictive algorithm that leverages artificial intelligence to find patterns across disparate data sources to determine which homes will likely go on the market in the next year,” Mintzer said.
Homesnap’s 60-person engineering team researched other platforms with seller prediction features and ran regressions, which allowed data scientists to determine accuracy based on the strength between those platforms’ dependent (e.g. the seller) and independent variables (e.g. age of home, homeownership tenure).
The results from the regressions enabled the team to fine tune the independent variables they tracked, which includes factors such as a seller’s homeownership tenure, a property’s last list date, the mortgage age, balance, and rate, and the scale of recent renovations.
“At its core, it’s really about pattern recognition,” Mazur explained.
In addition to considering seller and property factors, Homesnap also included a neural network feature that considers agents’ farming methods when creating predictions.
“What we found was that the more specific we were able to understand an agent’s farming strategy and the kinds of houses and clients they’re looking for, the more accurate we could be in determining potential sellers in the marketplace.”
“The neural network feature allows the algorithm to get better over time,” he added. “As the agent uses the platform more, the better the predictions get.”
Agents are then able to track homeowners and contact them within the Homesnap app as they move closer to listing.
“Ninety percent of sellers select the first agent they talk to,” Mintzer said. “Homesnap Pro agents can identify potential sellers before they’ve started the selling process, putting them one step ahead of the competition.”
Even though the Homesnap team is relishing today’s release of the All-New Homesnap Pro, they’re already planning launches for multiple upgrades that will be released throughout 2020.
“We want to make sure it’s rolled out in phases so agents have the predicate of, ‘I understand how to utilize the off-market marketplace,’ and once they have the command of the software, we’re going to roll out additional features that will make [All-New Homesnap Pro] more advanced.”