For CFOs, treasurers and underwriters

The warranty reserve is the last large liability still priced off a rear-view mirror.

US manufacturers held $71.89 billion in warranty reserves at the end of 2025, up 17% in a single year. Most of it is sized on historical loss data that no longer describes the product. iWarranty turns warranty into a real-time, structured data asset, the input that actually tightens the reserve.

→ Book a reserve impact reviewSee how it works
$30.37B
paid in US warranty claims in 2025, up 4%
Warranty Week, 23rd Annual Product Warranty Report
$71.89B
held in warranty reserves at year end, up 17%
Warranty Week
1.5–2.5%
of annual revenue lost to warranty cost
APQC
Auto
adjudication of a standard claim, no manual handoff
iWarranty

The cost isn't in a line item. It's in latency.

The real cost in warranty operations is the gap between something failing and a correct decision being made, and that gap leaks money across every line at once. The largest recoverable pool is leakage: the fraud, duplicate submissions and out-of-coverage payouts that slip through when adjudication defaults to rubber-stamping under volume. Across our deployment base we typically remove 12% to 18% of unrecoverable claims spend within two quarters. The speed of adjudication is only the visible part. The value is that every decision becomes structured data, which is what re-prices everything downstream.

When diagnosis gets cheap, the reserve changes.

Replacement became the default in electronics and appliances for one reason, and it was never cost: it was that diagnosis was expensive. AI collapses the cost of diagnosis, which pulls an entire band of claims back into the range where repair is the cheaper answer. A replacement charges the full cost of goods against the reserve; a repair charges a fraction, so the average cost per claim falls. It only reaches the reserve if the model is fed the new mix in real time, and most are not. They run on historical loss data that assumes the old replace-heavy world, which is why Warranty Week records billions every year in "changes of estimate", the polite term for reserves that turned out to be wrong.

Historical loss data is a rear-view mirror.

Reserve models assume failures arrive smoothly. They don't. Failures are front-loaded, with a sharp early-life spike, and a model built on annual averages misprices exactly the window where you most need to see clearly. The models break hardest in fast-refresh categories: consumer electronics, EV and battery, connected devices, where the historical data describes a product that no longer exists. Warranty Week reported Nvidia's warranty claims paid up 1,000% year on year in 2025, and US passenger-vehicle reserves up 25% to over $40 billion. Real-time warranty telemetry lets you reserve on leading indicators, claim velocity in the first weeks after sale, which is the difference between catching a reserve miss in week three and catching it a year too late. We examine the full reserve picture in our State of Warranty Intelligence 2026 report.

For OEMs

Compress claims-handling cost, defend your reserve assumptions with live data, and plan parts inventory against the actual repair-replace mix.

For insurers and extended-warranty providers

Price on leading indicators rather than lagging loss data, cut leakage, and reflect real regional risk instead of a single blended assumption.

See what real-time warranty data does to your reserve.

→ Book a reserve impact review

We use cookies to improve your experience and analyse site traffic. By clicking Accept, you consent to our use of cookies. Read our Privacy Policy and Cookie Policy for more information.