Riskified
E-commerce fraud detection platform vendor
Riskified is a company that provides a fraud detection and chargeback-guarantee platform for online retailers, evaluating card-not-present transactions in real time and approving or declining them on the merchant's behalf. Its core…
Definition
Riskified is a company that provides a fraud detection and chargeback-guarantee platform for online retailers, evaluating card-not-present transactions in real time and approving or declining them on the merchant's behalf. Its core commercial model transfers chargeback liability for approved transactions to Riskified in exchange for a fee, which distinguishes it from tools that only provide a risk score without taking on financial responsibility for the decision.
Overview
Riskified addresses the specific exposure online retailers face with card-not-present transactions, where the merchant, rather than the card issuer, typically bears liability for fraudulent chargebacks, creating pressure to decline any transaction that looks even mildly suspicious, at the cost of turning away legitimate revenue. Riskified's model directly targets that trade-off by guaranteeing the chargeback cost on transactions it approves, which shifts the merchant's incentive away from over-cautious manual review toward accepting more of Riskified's automated decisions. The platform evaluates each transaction using machine-learning models trained on a large volume of historical order data pooled across its retail customer base, examining signals such as device and browser fingerprints, shipping and billing address consistency, behavioral patterns during the shopping session, and known fraud indicators correlated across merchants, then returns an approve or decline recommendation typically within seconds so checkout is not delayed. Because Riskified is financially on the hook for approved orders that turn out fraudulent, its models are tuned with strong incentive alignment toward accuracy rather than simply toward flagging the maximum number of suspicious orders. Riskified is most often compared with Forter, the other major transaction-level e-commerce fraud vendor offering a liability-transfer commercial model; the two overlap substantially in function, with differences more pronounced in specific model tuning, industry vertical focus, and contractual terms than in fundamental architecture. Both differ from behavioral-biometrics vendors such as BioCatch, which detect account takeover during a session rather than evaluating a completed order, and from identity-verification vendors like Socure or Onfido, which authenticate who a person is rather than scoring an order's fraud risk. Online retailers, particularly in fashion, electronics, ticketing, and travel where fraud rates and average order values are both meaningfully high, adopt Riskified to reduce manual fraud review teams and to recover revenue previously lost to overly conservative decline rules. Retailers expanding into new international markets also use it to manage the added complexity of unfamiliar fraud patterns and payment methods without building in-house expertise. Some merchants specifically time a Riskified rollout to coincide with a large seasonal sales event, when order volume and fraud attempts both spike sharply and an in-house review team would otherwise struggle to scale fast enough to keep pace. Adoption considerations include the fee structure, typically a percentage of transaction value, which needs to be weighed against the specific merchant's existing fraud loss and decline rates, and the fact that liability-transfer guarantees are typically scoped to defined fraud types and transaction categories rather than covering every possible chargeback reason. Merchants with already low fraud rates may find the guarantee fee less cost-effective than an in-house or rules-based approach.
Key Features
- Chargeback liability transfer for transactions Riskified approves
- Machine-learning risk scoring using pooled historical order data
- Real-time approve/decline decisioning within checkout timeframes
- Device, address, and behavioral signal analysis per transaction
- Focus on card-not-present e-commerce fraud specifically
- International expansion support for unfamiliar fraud and payment patterns
- Reduced reliance on manual fraud review teams