Chargeback Guarantees
- A guarantee service approves or declines each order. Fraud on an approved order gets reimbursed. It's also incentivized to over-decline the borderline ones. That protects its margin, not yours
- No guarantee provider publishes a price. Signifyd publishes the model, not the number. It's a percentage of the order total on approved orders. Nothing gets charged when an order is declined for fraud. Riskified, Forter, NoFraud and ClearSale publish nothing at all (checked 2026-08-02)
- Compare covered losses and fees the guarantee could absorb with its full quote, declined-good margin and implementation cost, on the same sales population
- Test the guarantee against your current controls. Stripe Radar Lite is included with Stripe Payments; check the quote and your own contract before comparing costs
- Always negotiate a false positive SLA. Declined legitimate orders cost you money. They should own that too, not just fraud chargebacks
A guarantee vendor takes your fraud liability off you. They approve or decline each order, and when an approved order turns out to be fraud, they pay you back. You're buying certainty. The price of it is the good orders they decline to protect their own margin. Whether that trade is worth it comes down to one number. You can work it out in five minutes.
Do the math before signing anything. Use the number in your quote. Not one off a web page. The test is one line:
Buy only when covered losses and fees avoided, plus any recovered good-order margin, exceed the guarantee fee, added false declines and implementation cost. Compare the same period and eligible sales.
Take a 1% quote as an illustration. No vendor publishes a rate, so substitute yours:
- 0.5% fraud rate on $500K of revenue is $2,500 a year of fraud losses
- A guarantee quoted at 1% of approved orders costs roughly $5,000 a year
- You'd pay double your losses for the privilege
That illustration assumes all $500K is approved and billable, and compares only $2,500 of covered losses with the $5,000 fee. It leaves out dispute fees, changed approvals, implementation and blocked-good margin. Use the eligible approved-sales base in your quote, then price those other effects. Annual revenue alone doesn't settle the purchase.
Before spending money on a guarantee, fix your billing descriptor. Enable 3DS on high-risk segments. Both are free. Both hit the most common sources of disputes.
Chargeback guarantees are for card-not-present transactions. Card-present already has liability shift from EMV chip plus signature or PIN. CP-only? Skip this page.
How Chargeback Guarantees Work
Traditional fraud tools give you a score. "This order is 73% likely to be fraud." You still decide whether to ship. Get it wrong and you eat the loss.
Guarantee providers give you a decision: approve or decline. If they approve and it's fraud, they reimburse you.
Traditional: Score (73%) → You decide → You liable
Guarantee: Decision (Approve) → You ship → They liable
What's Covered
| Covered | Not Covered |
|---|---|
| Fraud chargebacks on approved orders | Friendly fraud (check your contract) |
| Chargeback fees | Non-fraud disputes (not as described, etc.) |
| Shipping costs (some providers) | Orders you approve against their decision |
| Chargebacks from transactions before contract |
Read your contract carefully. Coverage varies significantly between providers, and the friendly fraud line is the one to check.
Major Providers
| Provider | Model | Best For | What they publish about price (checked 2026-08-02) |
|---|---|---|---|
| Signifyd | Guarantee + score | Enterprise e-commerce | The model, not the number. "A percentage of the order total when an order is approved," varying by products, vertical, order volume and average ticket. "There is no charge when an order is declined due to fraud." (signifyd.com/pricing) |
| Forter | Decisions + guarantee options available | Enterprise, high volume | Nothing. No model, no tiers, no rate card - just "speak with one of our Forter experts" (forter.com) |
| Riskified | Guarantee | Fashion, luxury, high-AOV, travel | Nothing. The pricing page is a contact form (riskified.com/pricing) |
| ClearSale | Guarantee + manual review | International, emerging markets | Nothing |
| Kount (Equifax) | Score + optional guarantee | Flexible needs | Nothing. Note that kount.com now redirects to equifax.com/business/identity-fraud/, so you're buying an Equifax product line mid-rebrand |
| NoFraud | Guarantee | SMB e-commerce | Nothing |
Signifyd's "no charge when an order is declined due to fraud" is worth reading twice. The vendor only earns on orders it approves. That's better alignment than paying on total volume. Ask every other provider whether their fee works the same way. None of them says so publicly, which makes Signifyd's incentive the only one you can check before a sales call.
The Cost
Guarantee pricing works one of two ways:
| Model | How It Works | What vendors publish |
|---|---|---|
| Percentage of approved order value | X% of every order they approve | The percentage is not published by anyone. Signifyd confirms this is the model and that fraud declines carry no charge |
| Per-transaction fee | Flat fee per approved order | Not published. NoFraud describes a per-transaction model with the guarantee included, without a number |
So the honest cost table is blank. You can't fill it in without a quote. Run the arithmetic when the quote arrives.
Example math. The 1% below is an illustration, because nobody publishes a rate. Swap in your quoted percentage:
| Scenario | Guarantee Cost | Your Fraud Rate | Fraud Loss Without | Net Savings |
|---|---|---|---|---|
| $1M approved, quoted 1% | $10,000 | 0.5% | $5,000 | -$5,000 (lose money) |
| $1M approved, quoted 1% | $10,000 | 1.5% | $15,000 | +$5,000 (save money) |
| $1M approved, quoted 1% | $10,000 | 2.5% | $25,000 | +$15,000 (save money) |
The breakeven question: do covered losses and other measured benefits exceed the full added cost? The table isolates losses and fees on the same $1M base. It isn't a complete business case. Get the billing base, coverage and price in writing before a pilot.
One sanity check from the neighboring market. SEON sells scoring, not a guarantee, and it publishes $699/month for 2,500 fraud checks, roughly 28 cents a check (seon.io/pricing, checked 2026-08-02). It's the only public price in fraud tooling. A guarantee costs far more per order, and it should. It's buying liability transfer, not a score. But the gap can be enormous. If the vendor can't explain what they're absorbing, that's your answer.
The Hidden Cost: Declined Good Orders
Guarantee providers are incentivized to decline borderline orders.
They approve a $500 order and it's fraud. They lose $500. They decline it and it would have been good. You lose the sale. They lose nothing.
| Provider Incentive | Your Incentive |
|---|---|
| Minimize fraud payouts | Maximize approved revenue |
| Decline when uncertain | Approve when uncertain |
| Protect their margin | Grow your business |
The Forter argument: Forter now offers guarantees, but it historically pushed a technology-first approach instead. The argument: pure guarantee providers decline 5-15% of good orders to protect themselves. Better technology approves more orders and still covers the losses.
The Signifyd counter: guarantee providers approve more orders than merchants would alone. They have better data. So the net effect is positive.
Who's right? Test the provider on your own traffic. Compare covered losses and changed approvals with the full fee and blocked-good margin. A marketing argument or a fraud-rate cutoff won't settle your result.
When Guarantees Make Sense
Good Fit
| Situation | Why Guarantee Works |
|---|---|
| Covered losses and measured benefits exceed full added cost | The comparison includes fees, approvals and blocked-good margin on the same eligible sales |
| No fraud team | Outsource the expertise |
| High AOV | Single fraud loss is catastrophic |
| Approaching VAMP/ECM threshold | Can't afford more chargebacks |
| Low margins | Can't absorb fraud losses |
| Scaling fast | Don't have time to build fraud ops |
Poor Fit
| Situation | Why Guarantee Doesn't Work |
|---|---|
| Full added cost exceeds covered losses and measured benefits | The guarantee doesn't pay on the assumptions you've tested |
| Strong fraud team | Already doing well internally |
| Low AOV, high volume | Per-transaction fees add up |
| Mostly friendly fraud | Guarantees don't cover it |
| Tight margins | Fee exceeds fraud savings |
What to Ask Providers
Before signing:
| Question | Why It Matters |
|---|---|
| What's your approval rate for my vertical? | Tells you how many orders they'll decline |
| What's NOT covered? | Friendly fraud, specific reason codes |
| How fast do you reimburse? | Cash flow impact |
| Do you cover chargeback fees, not just the transaction? | Your processor's dispute fee is real money and it varies a lot. Verified on 2026-08-02: Stripe charges $15 to receive a dispute plus $15 to counter it, and returns neither on a partial win. PayPal charges $15.00 on wallet and Guest Checkout disputes or $20.00 on card chargebacks, never both, rising to a $30.00 High Volume Dispute fee under PayPal's eligibility test: its defined dispute ratio of at least 1.5% and more than 100 sales transactions in the previous three full calendar months, subject to published exclusions. Check the PayPal US agreement. Braintree charges $15.00. Moneris charges CA$25.00 plus a CA$80.00 authorization chargeback handling fee. Helcim charges $0 when you win and $15 when you lose. Square charges $0.00. The whole published US range is $0 to $30, so a guarantee sold on "we cover your $100 chargeback fee" is quoting a number no US processor publishes |
| What data do you need from me? | Integration complexity |
| Can I override your decisions? | Flexibility vs. coverage |
| What's the contract term? | Locked in for how long? |
| How do you handle disputes? | If you disagree with a denial |
Test to Run
Before committing to a guarantee:
- Measure eligible covered losses in dollars - Keep count rates separate, using the Fraud Metrics definitions
- Calculate your fraud cost - Fraud losses + chargeback fees + operational time
- Get guarantee pricing - Usually need to share volume for a quote
- Compare - Is guarantee cost < fraud cost?
- Ask about decline rate - What percentage of your orders would they decline?
The guarantee costs more than your fraud, and they'd decline orders you'd approve. You're paying to lose sales, so walk.
Scale Callout
Under $500K annual GMV: Start with current controls and compare a quote with actual eligible losses. A low sales total alone doesn't prove a guarantee can't pay.
$500K - $5M GMV: evaluate case by case. If fraud is a problem and you have nobody running it, a guarantee can be the right outsourcing call.
Over $5M GMV: Get competing quotes and compare outsourcing with the staff and tools needed internally. Better pricing has to appear in a written quote; sales volume doesn't promise it.
Where This Breaks
Friendly fraud dominant: most of your chargebacks are customers lying. They claim they never got the order, or that the charge was "unauthorized" when they placed it themselves. Guarantees won't help, because they cover true fraud, not customer abuse. See Friendly Fraud.
Mixed fraud types: check exactly which losses the contract covers. Compare covered amounts, fees and changed approvals with the full added cost. The larger fraud category by count doesn't decide whether the covered part is worth buying.
International complexity: some providers cover fewer countries. Others charge more for international transactions.
Alternatives to Guarantees
| Alternative | Pros | Cons |
|---|---|---|
| 3D Secure everywhere | Liability shift, no ongoing fee | Adds friction, some decline lift |
| In-house fraud team | Full control, no per-txn cost | Need expertise, slower to scale |
| Score-only tools | Cheaper than a guarantee, and at least one of them will tell you the price: SEON publishes $699/month for 2,500 checks | You keep the liability |
| Processor native tools | Already integrated, and free. Stripe Radar Lite is included with Stripe Payments, RevenueProtect with Adyen | Often less sophisticated |
| Manual review | Highest accuracy | Doesn't scale, slow |
See Fraud Vendors for the full landscape.
Next Steps
Evaluating guarantees?
- Calculate your fraud rate - Know your baseline
- Understand your fraud types - What's actually causing losses
- Get quotes from 2-3 providers - Compare pricing and coverage
Not ready for guarantee?
- Enable 3D Secure - Free liability shift
- Set up velocity rules - Catch obvious patterns
- Review processor tools - Use what you have
Already have high fraud?
- Survive a fraud attack playbook - Emergency response
- Check network program status - Know your thresholds
- Consider guarantee as bridge - While you fix root causes
Related Resources
- Fraud Vendors - Full vendor landscape
- Fraud Vendor Selection Guide - How to choose
- 3D Secure - Alternative liability shift
- Fraud Types - Understanding what you're preventing
- Fraud Metrics - Measuring your fraud rate
- Network Programs - VAMP, ECM thresholds
- Friendly Fraud - What guarantees don't cover
- Rules vs. ML - Detection approaches