Zillow was right to kill Climate RiskScores

Four row houses in South Philadelphia share a wall. Their flood risk scores: 6, 4, 1, 3.

This is real data from First Street, the climate analytics firm whose scores Zillow removed from over a million listings last month. The houses are on my block. They’re literally attached to each other, yet their risk profiles range from “minimal” to “major.” Clearly, a parcel-by-parcel scoring system isn’t ready for prime time when four houses get assigned wildly different scores like this.

More importantly, what does that climate risk score actually measure? First Street defines flood risk as “the likelihood of 1 inch of water reaching the building footprint of a home at least once within the next 30 years.” One inch. That’s a defensible threshold for probabilistic modeling, but it’s not what homebuyers think when they see “Major Flood Risk” next to a listing. They think Hurricane Helene, ruined basements, and destroyed HVAC systems. Perhaps they think twice about making an offer.

That would be troubling enough if buyers were carefully studying these scores. They’re not. Redfin, which has opted to keep climate scores on its listings, recently acknowledged that most shoppers rely on the site’s summary scores rather than clicking through to the full First Street report.

Inconsistent scores, consumers who only see a number, and a definition that bears no resemblance to ordinary understanding—this is precisely the kind of problem the legal system solved decades ago. When expert witnesses propose scientific evidence in litigation, courts apply what’s known as the Daubert standard: Is the methodology reliable? Does it help the jury decide the issue at hand? And, critically: does its potential to confuse or prejudice substantially outweigh its usefulness? Courts have long recognized that evidence wrapped in the authority of “science” can mislead even intelligent people when it’s oversimplified or misapplied.

The climate risk scores being placed on real estate listings fail this test. First Street’s models have legitimate applications for insurers and planners working at appropriate scales, but parcel-by-parcel data just isn’t reliable enough. Climate risk information can be relevant to consumers, but these scores so misalign with the common understanding of flood risk and damage that they raise anxiety more than inform—especially when we know that consumers are looking solely at that number like a Yelp rating, without even reading the review.

First Street’s CEO has acknowledged the interpretive challenge: “The complexity of physical climate risk models is difficult to understand. And if you’re not an industry expert, the nuances are hard to follow.” He’s right, and that’s precisely why displaying a simple score to non-experts is the wrong approach.

Bloomberg recently reported that two respected flood models (First Street’s and one from UC Irvine researchers) agreed on risk assessments just 21% of the time. When the scientists can’t agree, parcel-level precision is an illusion.

There’s a better path. Climate modeling should focus on broader community and regional assessments where the science is more robust. Flood risk should be defined in terms that connect to meaningful damage, not minimal exposure thresholds that trigger false alarms. Platforms that insist on showing these scores should make the methodology impossible to miss, not buried behind a click-through that Redfin acknowledges nobody uses.

Zillow made the right call. Removing the data doesn’t remove the risk. But displaying unreliable scores to consumers who won’t read the fine print, and using definitions that mislead, doesn’t advance climate literacy. It undermines it.

Anthony V. Mannino is the CEO of Dual Mind Strategies.
This column does not necessarily reflect the opinion of HousingWire’s editorial department and its owners. To contact the editor responsible for this piece: [email protected].