Opportunity

Southwest MLS (SWMLS), a leading MLS in the greater Albuquerque area, was looking to improve the quality of its listings. They had a robust process for many listing violations but had yet to find a consistent and timely way to validate that every property photo conformed to their standards. Due to 1,000’s of photos being uploaded every day, it was impossible for compliance staff to review every photo manually. As a result, they were forced to rely on agents to self-report violations, which resulted in non-compliant images slipping through the cracks and being syndicated downstream before being found.

Richard Gibbens, the MLS Director of SWMLS, was confident that AI, and specifically computer vision, could help ease the burden on his team and provide a more proactive solution.

Solution

SWMLS’s existing compliance solution, CoreLogic’s Listing Data Checker, had been integrating with Restb.ai’s photo compliance solutions for several years. Working closely with the Restb.ai team, they were able to configure their rules to tackle a seemingly simple, yet challenging, minimum photo requirement recently passed by SWMLS’s board.

The rule came about as Richard was aware some listings were uploaded that didn’t clearly show the entirety of the property. They considered a rule that simply relied on a minimum count of five photos, however, this didn’t solve the problem. Even if a listing had five photos, there was still no way to ensure that the main home spaces appeared in those images. As such, they decided to become the first MLS to require that every listing must have at least one photo of the front of the property, the kitchen, the bathroom, the living room, and the backyard/back of the property.

“Since AI can instantly scan all of our images, it allowed us to take a fresh look at what we wanted our listings to look like. We were no longer constrained by the time and effort it would require from our team to review images manually. Looking ahead, I see more MLSs coming up with new photo-based rules to ensure they have the highest quality of listings.”            Richard Gibbens – MLS Director at SWMLS

Impact

By working closely with the Restb.ai team, SWMLS was able to configure their LDC implementation to identify properties that lacked the minimum set of photos, alongside other violations, like images containing agent branding. Now, every time a listing is uploaded, they can instantly flag violations and minimize non-compliant listings from being syndicated downstream. 

20% of listing identified as sub-optimal

Even better, SWMLS easily and effortlessly benefits from AI’s ability to quickly learn and adapt. After Restb.ai reviewed the flagged images from SWMLS’s first 60 days of use, the algorithm was able to be retrained based on the particular issues identified in their region. Based on Restb.ai’s experience, it’s common for there to be unique violations in different markets. Furthermore, violations like agent branding can be subjective and what is acceptable may vary from MLS to MLS. Following this update, SWMLS saw a 49% reduction in the number of flagged violations that were deemed acceptable by their compliance staff.

40% reduction in violations

Most importantly, SWMLS was able to discover significant gaps in what images were being uploaded with each listing. They realized that while 99+% of listings contained photos of the front of home, kitchen and bathroom, up to 20% of properties didn’t have a clear photo of the property’s backyard and ~5% didn’t have an image of the living room. By providing SWMLS with these insights, they can now better educate their agents on listing requirements and improve the overall quality of their MLS’s data. 

Source: "Restb.ai"