Telematics for property: how insurers reward monitored risk

What auto telematics actually proved

The insight behind usage-based auto insurance was never the accelerometer. It was that a driver's own behavior, measured continuously, predicts loss better than the proxies underwriters had been using: age, ZIP code, credit, vehicle type. Once you can observe the risk directly, you stop pricing the category and start pricing the customer.

According to Progressive, its Snapshot program is "usage-based insurance" that "personalizes your car insurance rate based on your actual driving," measuring hard braking, late-night trips, mileage and, in some states, phone handling. Progressive reports that customers who renewed and earned a Snapshot discount saved an average of $328 a year, and says it has handed out over $2 billion in Snapshot discounts since 2009. Root goes further and makes the signal the product: according to Root, its app gives "a quote based primarily on how you drive," scoring safe hours, smooth braking, focused driving and gentle turning.

The part operators should notice is the direction of the incentive. The insurer is not charging for the data. It is paying for it, in discount dollars, because the observed risk is cheaper to underwrite than the assumed one. And drivers have largely accepted the trade: according to a January 2026 Insurance Journal report on survey work by the IoT Insurance Observatory and Arity, 82 percent of policyholders now view telematics apps positively, 60 percent are open to switching to usage-based insurance, and 52 percent would share a driving score to get personalized pricing.

Property insurance already runs this playbook

Homeowners carriers cannot put an accelerometer on a house, so they did the next thing: they bought the sensors themselves and gave them away. This is the same economics in different hardware. The carrier absorbs a device cost and a premium discount in exchange for a continuous signal on the two perils that dominate its loss ratio.

Those two perils are not a mystery. According to the Insurance Information Institute, water damage and freezing accounted for 27.6 percent of homeowners insurance claims in 2022, with average claim severity of $13,954 across 2018 to 2022, while fire and lightning made up 21.9 percent of claims at an average severity of $83,991, the most expensive category on the board. Frequency on one, severity on the other. If you only get to monitor two things in a building, those are the two.

The reason carriers can afford this is in the loss data. According to a LexisNexis Risk Solutions loss correlation study published May 5, 2020, homes that installed a Flo by Moen in-line water shutoff saw a 96 percent decrease in escape-of-water claim events in the year after installation, and 72 percent lower severity when a claim still occurred, while a comparison group without the device saw a 10 percent increase over the same window. The study covered 2,306 installed homes against a control group in the same geolocations.

Vendor-run numbers point the same way and should be read as vendor numbers: Whisker Labs claims Ting prevents "4 out of 5 electrical fires," and Nationwide's Phyn announcement cites an internal Phyn study of 13,500 homes over three years finding protected homes 99 percent less likely to have a non-weather water leak claim. We keep the compiled figures, methodologies and caveats on our water leak detection and insurance statistics page rather than restating them here.

The exterior is scored. The interior is not.

Sensors are only half of what underwriting already buys. The other half is computer vision, pointed at the outside of the building. According to Cape Analytics, its platform combines aerial and satellite imagery, weather data and public records to deliver "over 80 property insights at massive scale, ranging from current roof condition, roof age, and living area to hail and wildfire risk scores." Its July 2024 imagery collaboration with EagleView draws on an archive EagleView describes as more than three billion property photos covering 94 percent of the US population, with 20-plus years of history in some areas. ZestyAI runs the same play with named peril models, telling underwriters it detects "roof, vegetation, defensible space, secondary structures, slope, access, and more," with automated change detection over 20 years of historical imagery.

So an underwriter in 2026 can already answer, without visiting: how old is the roof, is it degrading, how much vegetation touches the structure, what is the wildfire and hail exposure. What no vendor in that stack can answer is anything happening on the other side of the walls.

This is not a criticism of the sensor vendors. Water and arcing are the right two things to instrument, and they are instrumented well. It is a statement about coverage. A leak sensor sees a leak once the water is already moving. It never saw the stain that spread across the ceiling for six weeks first, and it will never see the failed grout that caused it.

The visual sensor already exists in operated properties

Here is the asymmetry that makes commercially operated property different from an owner-occupied home. A vacation rental, a hotel room, and increasingly a multifamily unit is photographed on a schedule, by someone whose job it is, at a known moment in the property's life. A cleaner shoots the room after every guest. An inspector shoots it before the next check-in. That documentation already exists, it already happens hundreds of times a year per unit, and nobody built it for underwriting.

Structurally, it is telematics. A continuous behavioral record produced by normal operation, timestamped, tied to a specific asset, at a cadence no annual inspection can match. The difference is that the driving score needed a device installed to exist, and the condition record does not. It is already being captured and then almost entirely ignored, because no human can review a hundred photos per turnover across hundreds of units.

That last point is the one carriers should care about most, and it is the one no imagery vendor can supply. Two identical buildings with identical roofs are not identical risks if one operator documents and audits every turnover and the other does not. Until now there was no way to see that difference from the outside. A condition record makes an operator's discipline legible, in the same way a driving score made a careful driver legible.

This is the part RapidEye works on, and we will be direct about it rather than pretend otherwise. We analyze the photos and video that operators already capture between guests, compare each turnover against a per-room baseline, and flag what changed. In a trial with a 500-plus unit manager we analyzed over 1.5 million existing photos and surfaced an average of four damages per property that cleaners and inspectors had passed over. No hardware, no site visits, no new behavior asked of the cleaning team. Our page for insurance providers lays out what a continuous condition record could mean on the underwriting side, and we work the same data into damage waiver and protection programs, where evidence quality decides whether a claim is payable. The mechanics of using timestamped turnover documentation to establish exactly when damage appeared are covered in our piece on proving which guest caused the damage.

The operator's side of the trade

Telematics has a cautionary chapter, and property operators should read it before handing anyone a data feed. On January 16, 2025, the Federal Trade Commission took action against General Motors and OnStar over allegations they "collected, used, and sold drivers' precise geolocation data and driving behavior information from millions of vehicles" without adequately notifying consumers or obtaining affirmative consent. According to the FTC's complaint, that data was sold to consumer reporting agencies, which compiled it into reports insurers used to deny coverage and set rates, and some consumers did not know they had been enrolled at all. The proposed order banned GM and OnStar from disclosing that data to consumer reporting agencies for five years.

Nothing about that case argues against monitored risk. Progressive's discount and State Farm's free sensor are both monitored risk, and both are broadly welcomed, because the customer knew, chose, and got something. What GM got wrong was the consent, and the fact that the flow ran one direction. So if condition data is going to move from an operator to an insurer, the terms are worth stating up front.

Handled that way, the incentives line up cleanly. The operator was going to photograph the property anyway. Reviewing those photos catches damage while it is a maintenance ticket instead of a claim, which is good for the operator's margin before an insurer is involved at all. The carrier's interest in that same record is a second-order benefit, and it should stay second-order. That ordering is what keeps this from turning into the thing telematics almost became.

Quick FAQ