Riskified vs Sift (2026)
The choice is between a fee that moves with your approved sales and a platform you license by the year. Riskified takes a percentage of the value of each order it approves, and a declined order costs nothing. When fraud gets through on an approved order, Riskified reimburses the chargeback. Its Dispute Resolve evidence service comes with that guarantee at no added cost. Sift's one public price is for platform access in 12 month contract units, and its terms leave the loss from a bad call with you. For chargebacks, Sift sells Dispute Management to teams that contest their own. The scope differs as well. Riskified is built around the online order, with separate products for refund abuse, logins and checkout friction. Sift scores account and payment events for consumer apps as well as stores. Its industry pages run from ecommerce and travel to iGaming and fintech. Sift also hands your team the levers. RiskWatch can move block thresholds during an attack and bring them back afterward. Pick Riskified to tie fraud cost to revenue on retail, travel or ticket orders. Pick Sift to tune and defend accounts and payments across an app yourself.
- You want the vendor paying for its own misses. Riskified reimburses fraud chargebacks on the orders it approves, and the fee it earns rises only when your sales go through.
- AI shopping agents are reaching your store. Riskified lets merchants set rules by type of agent, approving low risk agent orders and sending high risk refund claims to review, and agents can query it through an MCP server on AWS Marketplace.
- You want a say on borderline orders. Riskified's Recover gives merchants more control over the final call on them, and its ARIA analyst explains decisions at the level of each transaction.
- Your customers shop from many countries. Riskified works from New York, Tel Aviv, London and Shanghai, and five of its ten largest new customers in mid 2026 were based outside the United States.
- You sell travel or event tickets as well as goods. Riskified names Air Europa and Lastminute.com in travel and Gametime in tickets, and Finish Line credits it with a 70 percent drop in chargebacks.
- Fraud hits you in bursts. Sift's Fraud Attack Defense Suite, launched in October 2026, spots traffic spikes and card testing and pairs each alert with a playbook. A person closes each alert, and the history stays searchable for audit.
- You run a fintech or consumer app rather than a store. Sift's fintech page covers neobanks, crypto exchanges and digital wallets, with figures from doxo and CoinJar.
- Your team already contests its own chargebacks. Sift Dispute Management builds and sends the responses, and it traces to Chargeback, a dispute company Sift agreed to buy in 2021.
- Your data team wants models that know your market. Sift blends a global model with cohort models that learn from one industry or region and a custom model trained on your own outcomes.
- Chargeback rate is the number you answer for. ChowNow reports a 99 percent drop in its chargeback rate on Sift and an average fraud rate under 0.03 percent.
This comparison is published by AI FinTech Index, an independent research platform that publishes independent ratings of AI vendors for financial services. Riskified and Sift are each graded against the same capability taxonomy, from each vendor's own public materials and the regulatory record, under the AI FinTech Index verification standard. No vendor pays for placement, and no vendor has reviewed this page. How this evidence is graded
Plain facts
| Riskified | Sift | |
|---|---|---|
| Primary category | Fraud Detection & Transaction Risk | Fraud Detection & Transaction Risk |
| Founded | 2013 | 2011 |
| Headquarters | New York, New York, United States | San Francisco, California, United States |
| Website | www.riskified.com | sift.com |
Side by Side
Select any grade to read the note behind it.
| Axis | R Riskified |
S Sift |
|---|---|---|
| AI Centrality | ||
| Autonomy and Oversight Model | ||
| Model Risk Management and Transparency | ||
| Operational and Outcome Evidence | ||
| AI Safety and Data Stewardship | ||
| GLBA and Data Privacy Posture | ||
| Security Certifications and Trust Center | ||
| Regulatory Status and Licensure | ||
| AI Governance and Bias Disclosure | ||
| AI Liability and Recourse | ||
| Model Supply Chain Disclosure | ||
| Core Systems and Integration Depth | ||
| Deployment Model and Data Residency | ||
| Commercial Transparency | ||
| Institution and Segment Coverage |
Product updates
The latest logged changes from each vendor, newest first, each linked to its source.
Riskified
-
Integration / interoperability
Riskified brought its Identity Risk Intelligence into Zendesk Industry Agents.
Source
Sift
-
Product / capability
Sift released Review Agent, which reads a fraud case and gives the analyst color coded risk and trust signals, a short plain language explanation and a suggested action of Block, Accept or Watch.
Source
The short version of each
Riskified
Riskified sells online merchants a decision on every order, approve or decline, with its own money behind each approval. Retailers, travel firms and ticket sellers pay a share of the value of the orders it approves, so the fee grows only when sales go through. Fraud chargebacks on approved orders are paid back under its Chargeback Guarantee, and Riskified carries at least $10 million of professional liability cover. Policy Protect handles refund abuse, Account Secure guards logins and Adaptive Checkout matches friction to risk. Native connectors reach Shopify, VTEX and commercetools. Card issuers on Marqeta can now draw on Riskified's risk data at authorization, and its payment coverage reaches past cards to ACH. Kogan reported $1.5 million in yearly savings in a July 2026 release. Riskified Ltd. is incorporated in Israel and trades on the New York Stock Exchange. The main caveat is how much a merchant can inspect. Riskified publishes no decline threshold, error rate or fairness test, and its subprocessor list, data processing agreement and audit reports come only after an access request.
Source: AI FinTech Index, 2026
Sift
Sift gives fraud teams a risk score on each sign up, login, payment and dispute event a business sends, and leaves the call to them. Founded in 2011 and privately held, it serves digital consumer businesses and fintechs, with ChowNow, doxo and CoinJar among the customers it names. Each score blends a global model, cohort models for an industry or region and a custom model for the customer. Teams connect through a public API, a JavaScript snippet and mobile SDKs, and a read only MCP server lets their own AI tools query Sift. Dispute Management drafts and submits chargeback responses. A cross functional Ethics Committee oversees the models under a governance policy Sift aligns with Annex B of ISO 42001. Its public data protection addendum gives customers 30 days to object to a new subprocessor. The one published price is $10,000 per 12 month contract unit on AWS Marketplace. The main caveat is who pays for mistakes. Sift's terms promise no accurate results, make the customer solely responsible for its decisions and cap Sift's liability at a year of fees.
Source: AI FinTech Index, 2026
Common questions
How does Riskified compare with Sift?
Riskified and Sift both score online activity in real time on data pooled across their customers, but they sell different things. Riskified sells the decision on each online order. It charges a percentage of approved order value and reimburses fraud chargebacks on the orders it approves. Sift sells a scoring platform, and the customer's own workflows decide what to block, accept or send to review. Riskified's buyers are online retailers, travel firms and ticket sellers. Sift serves stores too, and its industry pages also reach travel, iGaming and fintech apps. A merchant that wants fraud cost tied to revenue leans toward Riskified. A team that wants to tune its own defenses across an app leans toward Sift. Graded by AI FinTech Index against the same capability axes from each vendor's own published materials, verified November 5, 2026. No vendor pays for placement.
What do Riskified and Sift cost?
Neither Riskified nor Sift publishes a rate card. Riskified writes a percentage of approved order value into each merchant's order form, set by category, order values, geography and fraud profile, and charges nothing on declined orders. Dispute Resolve evidence collection is included with its guarantee. Sift's one public price sits on its AWS Marketplace listing for Dispute Management, at $10,000 per 12 month contract unit, and the listing does not say what a unit covers. Custom pricing runs through private offers. A Sift trust center answer ties console single sign on to plan type, though no plan is described in public. Graded by AI FinTech Index against the same capability axes from each vendor's own published materials, verified November 5, 2026. No vendor pays for placement.
How do Riskified and Sift handle chargebacks and disputes?
Riskified takes the fraud chargebacks itself. Its contract pays each one on an approved order up to the order value, while the merchant's fees are current. Connectors to Adyen, Braintree, PayPal and Stripe reconcile disputes against the payment layer, and Dispute Resolve gathers the evidence at no added cost. Sift leaves the loss with the customer. Its terms make the customer solely responsible for its decisions and cap Sift's liability at a year of fees. Sift instead sells the tools to contest chargebacks. Its Dispute Management product builds and sends the responses, and it runs on Amazon Web Services in Oregon, apart from the rest of Sift on Google Cloud. Graded by AI FinTech Index against the same capability axes from each vendor's own published materials, verified November 5, 2026. No vendor pays for placement.
Do Riskified and Sift cover refund abuse and account takeover as well as card fraud?
Riskified sells a separate product for each problem. Policy Protect targets refund abuse, and Account Secure covers logins and is sold as fully automated. Adaptive Checkout adds friction in proportion to an order's risk. Sift's platform pages lead with two products. Payment Protection screens for stolen cards, card testing and chargeback fraud, and Account Defense watches for account takeover and fake accounts. Account Defense acts on one session rather than the whole account, and Payment Protection can add a check in place of a decline. Sift's platform pages present no refund abuse product, so a merchant whose losses come from returns has that question to put to Sift. Graded by AI FinTech Index against the same capability axes from each vendor's own published materials, verified November 5, 2026. No vendor pays for placement.
How do Riskified and Sift handle card data, security and European payment rules?
Riskified says it is not required to follow PCI DSS because it never collects full card numbers. Sift names no PCI DSS certification. Riskified's trust center lists ISO/IEC 27001, 27017 and 27018 with SOC 1, SOC 2 and SOC 3 reports, and Sift's lists ISO 27001:2022 and a SOC 2 Type II report, each behind an access request. Sift encrypts data with AES 256 at rest. For Europe, Riskified's transaction risk analysis exempts low risk orders from strong customer authentication, and KIKO Milano ran that exemption through the payments hub Axerve. According to the AI FinTech Index, Sift's Payment Protection claims PSD2 compatibility but names no exemption it supports. Riskified may store and process data in the United States, the EU and Israel, and its trust center names Amazon Web Services. Sift hosts in the United States on Google Cloud and says data may also be processed in Ukraine and other countries.
Related comparisons
Other published head to head assessments involving these vendors or their closest peers. The full set for this category is on the Fraud Detection & Transaction Risk page.
Neither Riskified nor Sift publishes error rates, false decline rates or a validation report, and neither shows how its decisions fall across groups of shoppers. Riskified's 20-F concedes that wrong merchant data can link a legitimate order to fraudulent ones, yet no fairness testing or model approval process is published. Sift names an Ethics Committee and a governance policy and shares no result from either. Both pool shopper data across customers, and neither describes a way to opt out.
Neither has a financial license, which suits a technology supplier, and Riskified's 20-F lists emerging rules for AI among its risks. Sift's addendum names Google as the generative AI supplier for its session summaries, while Riskified names no model provider for ARIA. Riskified owes written answers on three points.
What rate would a merchant of your category and order profile pay, who sits on its subprocessor list, and does a services privacy policy dated September 2021 still describe the service? Sift owes written answers on what one AWS Marketplace unit covers, which model runs its Review Agent, and whether its scores reach refund abuse.