Riskified is a publicly traded company that provides software for ecommerce risk management, including fraud prevention and chargeback management. Its platform is designed to help online merchants approve legitimate transactions while identifying and blocking fraudulent activity, with a focus on reducing the financial and operational burden of chargebacks. The company's technology is built around Machine learning models and behavioral analysis, offering a guarantee that approved transactions are protected against fraud-related chargebacks.
Founded in 2012 by Eido Gal and Assaf Feldman, Riskified has grown from a startup into a major player in the ecommerce risk management sector. The company's core value proposition is its chargeback 100% money-back guarantee: if a transaction approved by its technology results in a fraud chargeback, Riskified reimburses the merchant. This model shifts the risk from the merchant to Riskified, incentivizing the company to continuously refine its detection algorithms.
History
Riskified was founded in 2012 by Eido Gal, who serves as CEO, and Assaf Feldman, who serves as CTO. The company was established in Tel Aviv, Israel, with the aim of addressing the growing challenge of fraud in online retail. Early on, the founders recognized that traditional rule-based fraud detection systems were inadequate for the scale and sophistication of modern ecommerce fraud, leading them to develop a Deep learning-based approach.
By 2018, Riskified had secured a total of $63.7 million in funding from various investors. This capital was used to expand its engineering team and enhance its Neural network models, which analyze thousands of data points per transaction, including device fingerprints, shipping addresses, and purchasing patterns. In November 2019, the company announced a Series E funding round of $165 million, led by General Atlantic and joined by Fidelity Management & Research, Winslow Capital, and existing investors. This round brought the company's total funding to over $228 million and was intended to support international expansion and product development.
On July 28, 2021, Riskified launched its initial public offering on the New York Stock Exchange under the ticker symbol RSKD. The IPO valued the company at $4.3 billion, reflecting strong investor confidence in the ecommerce fraud prevention market. Following the IPO, the company continued to invest in Artificial intelligence research, particularly in areas like Transformer (architecture) architectures and Large language model applications for fraud detection.
Technology and Approach
Riskified's technology relies on a combination of behavioral analysis, elastic linking, proxy detection, and machine learning. Behavioral analysis examines how users interact with a website, such as mouse movements, typing speed, and navigation patterns, to identify anomalies that may indicate automated bots or fraudulent intent. Elastic linking is a technique that connects seemingly unrelated data points across transactions to uncover networks of fraudsters, while proxy detection identifies users who are hiding their true IP addresses.
The core of the platform is a suite of machine learning models that are trained on billions of historical transactions. These models use Residual Network (ResNet) and Batch Normalization techniques to improve accuracy and reduce false positives. Unlike traditional systems that rely on static rules, Riskified's models continuously learn from new data, adapting to emerging fraud patterns. The company also employs Data Augmentation to generate synthetic training examples, improving the robustness of its models against novel attack vectors.
Riskified's system is designed to make real-time decisions, typically in under 200 milliseconds, allowing merchants to provide a seamless checkout experience. The platform integrates with major ecommerce platforms and payment gateways, such as Shopify, Magento, and Stripe, via APIs. This integration enables merchants to automate the approval or rejection of orders without manual review, significantly reducing operational costs.
Products and Services
Riskified offers a suite of products tailored to different aspects of ecommerce risk management. The primary product is its fraud prevention platform, which evaluates transactions at the point of sale and provides a decision to approve, decline, or review. This product is complemented by chargeback management services, which help merchants dispute fraudulent chargebacks and recover revenue.
In addition to fraud prevention, Riskified provides tools for policy compliance and account security. For example, its platform can detect account takeover attempts, where fraudsters use stolen credentials to make unauthorized purchases. The company also offers a solution for managing "friendly fraud," where legitimate cardholders falsely claim that a transaction was unauthorized. By analyzing customer behavior and order history, Riskified can identify patterns indicative of friendly fraud and provide evidence for chargeback disputes.
Riskified's services are used by a range of merchants, from small online retailers to large enterprises. The company reports that its technology has analyzed over 5 billion transactions since its founding, with a fraud detection accuracy rate that it claims exceeds 99%. The chargeback guarantee is a key differentiator, as it aligns Riskified's incentives with those of its clients: the company only profits when it correctly identifies legitimate transactions.
Market and Competition
The ecommerce fraud prevention market is highly competitive, with players ranging from traditional payment processors to specialized AI startups. Riskified competes with companies like Forter, Signifyd, and Kount, which offer similar machine-learning-based solutions. However, Riskified distinguishes itself through its focus on the chargeback guarantee and its deep integration with enterprise-level merchants.
The market has grown significantly due to the rise of online shopping and the increasing sophistication of fraudsters. According to industry reports, ecommerce fraud losses are projected to exceed $48 billion annually by 2023, driving demand for advanced prevention tools. Riskified's use of Generative AI and Sequence-to-Sequence (Seq2Seq) models has been noted as a competitive advantage, as these techniques allow for more nuanced analysis of transaction sequences and user behavior.
The company also faces competition from in-house fraud teams at large retailers, which may develop proprietary systems. However, Riskified argues that its cross-industry data, drawn from thousands of merchants, provides a broader view of fraud patterns than any single merchant could achieve. This network effect is a key part of its value proposition.
Corporate Structure and Leadership
Riskified is headquartered in New York City, with its primary research and development center in Tel Aviv. The company has additional offices in London and Shanghai, reflecting its global client base. As of 2024, Riskified employs over 600 people, with a significant portion of its workforce dedicated to data science and engineering.
Eido Gal, co-founder and CEO, has been the public face of the company, frequently speaking at industry conferences about the future of ecommerce risk management. Assaf Feldman, co-founder and CTO, oversees the technical direction, including the development of proprietary algorithms and infrastructure. The company's board includes representatives from major investors like General Atlantic and Fidelity.
Riskified has also invested in academic partnerships, collaborating with institutions such as University of Toronto and Carnegie Mellon University to advance research in Deep learning and fraud detection. These partnerships help the company stay at the forefront of AI innovation and attract top talent.
Financial Performance and Growth
Following its IPO, Riskified has reported steady revenue growth, though it has also incurred net losses due to heavy investment in research and development. In 2022, the company reported annual revenue of $307 million, a 30% increase from the previous year. However, net losses widened to $120 million as the company expanded its sales team and invested in new product lines.
In 2023, Riskified announced a partnership with Amazon Web Services to leverage cloud infrastructure for scaling its machine learning models. This collaboration aims to reduce latency and improve the cost efficiency of its transaction processing. The company has also explored the use of AWS Trainium chips for training its models, which could reduce compute costs by up to 40%.
Riskified's customer retention rate is high, with over 90% of its top 100 clients renewing their contracts annually. The company has expanded into new verticals, including travel, digital goods, and subscription services, which have different fraud profiles than traditional retail. This diversification has helped mitigate the impact of economic downturns on consumer spending.
Regulatory and Ethical Considerations
As a company handling sensitive financial data, Riskified is subject to regulations such as the Payment Card Industry Data Security Standard (PCI DSS) and the General Data Protection Regulation (GDPR) in Europe. The company maintains compliance with these standards through regular audits and data encryption practices. Its use of behavioral data has raised privacy concerns, but Riskified states that it anonymizes data and does not sell personal information to third parties.
The company has also been proactive in addressing algorithmic bias, ensuring that its models do not unfairly discriminate against certain groups of consumers. Riskified publishes transparency reports and has an internal ethics committee that reviews new AI features. In 2023, it joined the BAIR (Berkeley AI Research) consortium to collaborate on responsible AI practices.
Future Outlook
Riskified continues to invest in emerging technologies, particularly Large language models, to improve its fraud detection capabilities. The company is exploring how Transformer (architecture)-based models can better understand the context of transactions, such as natural language in customer reviews or support tickets, to identify fraud signals. It is also developing tools for merchants to visualize and understand fraud patterns through interactive dashboards.
The ecommerce fraud prevention market is expected to grow at a compound annual growth rate of 20% through 2030, driven by the expansion of cross-border ecommerce and the adoption of new payment methods like buy-now-pay-later. Riskified is well-positioned to capitalize on this growth, given its established brand and technological edge. However, it faces challenges from increasing regulatory scrutiny and the need to continuously innovate to stay ahead of fraudsters.
In summary, Riskified has established itself as a leader in AI-powered ecommerce risk management, with a unique business model that aligns its success with that of its clients. Its commitment to research and development, combined with a strong financial foundation, suggests that it will remain a significant player in the industry for years to come.