Extended warranties on a television. Screen cover on a phone. The finance desk's menu after you have agreed the price of a car. Credit life on a loan. Travel cover at the checkout of a booking. A home warranty on a house you have just bought.
They cover completely different things. What unites them is the moment: you are sold them after you have already decided to buy the main thing, by someone paid out of the price, at a point where walking away to compare costs you the whole transaction.
That moment is the product. Everything else follows from it.
Insurance has an objective measure: the share of premium that comes back out as claims. It does not tell you whether a product is right for you, but it tells you how much of your money is buying protection and how much is buying distribution.
American figures come from different regulators using different methods, so the third column matters as much as the second. Some are measured against the premium that reached an insurer, after the seller's cut was already taken out. Only one is measured against what the customer actually handed over.
| Cover | Paid back as claims | Measured against |
|---|---|---|
| Warranty line, all writers in one state | 77% | premium reaching the insurer16 |
| Homeowners insurance, ten-year average | 63% | premium reaching the insurer17 |
| Credit life, one state | 45% | premium reaching the insurer18 |
| Vehicle service contracts, one company | 42% | what the customer paid19 |
| Home warranty, one state | 36% | premium reaching the insurer20 |
| Consumer product protection plans, one seller | 25% | premium reaching the insurer21 |
| GAP | nobody publishes it | — |
Two things in that table are worth more than the individual numbers.
The first is the third column. A figure measured against premium reaching the insurer has already had the retailer's or dealer's cut removed before the arithmetic starts. So the share of your money that came back is lower than the number shown — by an amount nobody publishes. Only the vehicle service contract row measures the whole dollar you paid, and it is the one to compare against.
The second is the last row.
GAP is the clearest case in this chapter, and not because it pays out badly. Because in the United States you cannot find out what it pays out at all.
No state publishes it. No federal agency publishes it. The industry does not publish it. The entire American record is one underwriter's own filing showing a 20.9% loss ratio for half of 2010 against 90.4% the year before — one company, one half-year, sixteen years ago, swinging so far between adjacent years that it can stand for nothing.22
There is a reason for the silence. In much of the country GAP is not sold as insurance at all. It is a debt waiver — the lender agreeing to forgive the shortfall — and a debt waiver never enters the data insurance departments collect. The money is real and the reporting category does not exist.
One state has set a standard rather than a measurement: Oregon requires filed GAP insurance rates to anticipate that at least 50% of premium comes back as claims.23 That is a bar for approving a rate. It is not a record of what happened, and no state publishes that.
What is measured is the price. The Consumer Financial Protection Bureau, working from loan-level data on roughly 34 million auto loan originations, put the average cost of GAP financed into a US auto loan at $952.24 That sits above the $400 to $700 that consumer articles usually quote.
And almost nobody pays cash for it. Financed at the rates prevailing in early 2026, that $952 costs about $1,123 on a typical new car loan and about $1,296 on a used one — roughly a third more than the number on the form.25
So at the GAP line specifically you are asked to accept a price of about a thousand dollars, which is really thirteen hundred, for a product whose value nobody in the country measures.
There is a puzzle here worth resolving, because it explains the whole family.
If you look up the warranty business on US insurers' books, it does not look predatory at all. The line paid out 66.0% of premium in claims in 2023.3 That is unremarkable — close to home insurance.
And yet the consumer figure is nothing like that. Consumer Reports' guidance on service plans is blunt about why: retailers push them "because they're cash cows for them. Stores keep 50 percent or more of what they charge for these contracts."4
Both are true at once, and together they explain everything. The retailer's share leaves before your money reaches the regulated insurer. A 66% loss ratio on a premium that is half of what you paid is around a third of what you paid. The insurance sitting behind the counter is priced normally. The thing you bought is not the insurance — it is the insurance plus a markup that never appears on any statement.
This is why "but the insurer isn't making much" is not a defense, and why loss ratios published at the wholesale layer will always look better than what you experienced.
The best research on this used four years of real till data from an electronics retailer. Two findings matter.5
Where people buy tells you it is the moment, not the product. Attachment was 28.7% in store against roughly 4% online — for the same goods, at the same prices, from the same retailer. Nothing about the risk changed between the two. Only the presence of a person asking.
People are not buying insurance, they are buying a mistaken belief. Buyers systematically overestimate the chance of failure — a 5% real probability was perceived as about 13%. And when researchers simply showed shoppers the actual failure rates, what they were willing to pay dropped significantly.
That last result is unusual and worth sitting with. Most sales resistance requires willpower. This one requires a number. The product survives on the gap between the failure rate people imagine and the failure rate that exists.
A car dealership makes real money in the finance office, after the price of the car is settled. Across publicly traded US dealer groups, gross profit from finance and insurance ran $2,505 per vehicle in the first quarter of 2025, $2,515 in the second and $2,534 in the third — up 5.2% on the year, driven by selling more products per deal rather than by charging more for each.6
That is the average, on every car. It is why the menu appears after you have agreed a price and shaken hands, and why the conversation is unhurried in a way the price negotiation was not.
Regulators have been active here. In 2023 a major auto lender was ordered to pay $60 million — $48 million in redress and a $12 million penalty — including nearly $32 million to customers who were denied refunds of unearned GAP and credit insurance premiums. In that case the bundled products averaged $700 to $2,500 per loan.7
A rule that would have required dealers to disclose add-on prices up front was struck down by a federal appeals court in January 2025 and is not in force.8 Anyone writing about this needs to check its status again before relying on it either way.
If this chapter only said no, it would be as useless as the sales pitch. Three of these products transfer a real, concentrated loss that a lot of households genuinely cannot absorb.
GAP. If your car is written off, the insurer pays what it is worth, and you still owe whatever the loan says. That difference is a real, undiversifiable loss on a car you no longer have. The product is right for someone who could not write that check.
And that group is large and growing. The share of used-car loans written for more than 120% of the car's value has gone from about a third before the pandemic to 54% — and the share written above 140% of the value has gone from 13% to 32%.14 A third of used-car borrowers now owe half again what the car is worth on the day they drive it away. For them the gap is not a remote possibility; it is the starting position.
The dealer's version is a different question. No American regulator publishes what GAP returns through any channel, so nobody can tell you the desk is worse value than your own insurer — only that the desk's price is measured and high. Get a quote from your own carrier before you decide; it takes ten minutes and it is the only comparison anyone can actually make.
Travel medical and evacuation. An overseas hospital bill and an air ambulance have no ceiling, and a US health plan often will not pay them. This is the one product in the family that passes the test the others fail: the loss it covers could genuinely ruin you, and nothing else you own covers it. Note the shape of it: the case is for the medical and evacuation piece, not for cancellation cover on a cheap domestic flight.
A service contract on a known expensive failure. A model with a documented, common, costly failure — a transmission, a hybrid battery — is a different risk from a television.
And the honest anomaly, which matters more than any of them:
For a household with no cushion at all, a plan that is bad value on average can still be a reasonable purchase. A $600 repair bill that means a payday loan is not the same event as a $600 repair bill. Buying a bad deal to avoid a worse one is a rational trade, not a failure of financial literacy — and anyone lecturing about expected value to someone with $40 in the account has misunderstood the situation.
Before buying any of this, two checks that cost nothing.
The manufacturer's warranty already covers the first stretch, and an extended plan often runs alongside it rather than after it — so part of what you buy may be time you already have.
Your credit card may extend it for you. Card extended-warranty benefits typically add about a year, with caps commonly around $10,000 per claim and $50,000 per account. They are not universal — some major issuers have dropped the benefit entirely.9 It is worth knowing which card you used.
This is the least-known useful fact in the whole chapter. These products are generally cancellable, often with a full refund inside a window and a pro-rata refund after it.
The details are set by state law and vary, but the shape is consistent:
vehicle without a manufacturer warranty, with pro-rata refunds after that less a fee capped at the lesser of 10% or $25.10
mandates a minimum 30-day free look, and adds a 10% penalty on refunds not paid within 30 days.11
allows cancellation within 30 days for any reason or none, with a full refund where no claim has occurred.12
There is also a refund most people never claim: if you sell the car, refinance, or pay the loan off early, the unused portion of a GAP or credit insurance premium is generally owed back to you. The federal consumer regulator states plainly that you can cancel add-on products at any time.13 Nobody sends a reminder.
Two, in this order.
If this went wrong and I had no cover, could I write the check? Not "would it hurt" — would my life change. If you could pay it from savings, the premium is buying convenience, and convenience at these payout rates is expensive.
How much of what I pay comes back? If you cannot find out, that is itself the answer.
Everything else follows. Fail the first and do not buy it at any price. Pass the first but fail the second and the answer is buy it — somewhere that is not this desk.
These purchases are invisible in a way few others are. They arrive bundled into a loan payment, a monthly line on a phone bill, a single charge inside a larger sale. Plenee's job is to surface them as what they are: a set of small recurring payments that, added up across a year, are usually larger than anyone guesses — and several of which can be canceled this week for a refund.
These products are defined by when they are sold, not by what they cover. The published payout rates run from about three-quarters of premium down to a quarter — and for GAP, the product sold hardest at the car desk, no American regulator publishes a figure at all. The insurer's numbers look better than they are because the retailer's cut left before the money got there. People buy in store at seven times the rate they buy online, and buy less when shown the real failure odds. Three of these — GAP, travel medical, and a service contract on a known bad mechanism — are genuinely worth having, though rarely at the desk offering them. And most of them can be canceled for a refund you have to ask for.
Plenee Academy provides financial information and education, not personalized financial advice. Plenee Co. is not a registered investment adviser, broker-dealer, or financial planner. Legal Disclosures & Notices →