Founded in 1995, Getty Images has proven expertise in the editorial space, including visual coverage of current events, entertainment, and sports. It leads in the stock imagery industry, and it has the largest privately owned archive in the world. Getty Images was the first company to license imagery on the web but eventually moved the industry online.
In 2006, Getty Images acquired iStock, the original source for user-generated stock photos, vectors and illustrations, and video clips. Getty Images and iStock are a force in the broader visual content industry. They give photographers and visual artists a platform to license content to media, designers, freelancers, small-business owners, agencies, and more.
As Getty Images moved online, it focused on providing content to the world. But it faced an inevitable challenge: eCommerce payments fraud and the resulting chargebacks. In 2020, Aite Group spoke with Getty Images and iStock to understand how they addressed their challenges using the Kount Command eCommerce fraud prevention solution.
Typically, financial institutions are responsible for card-present fraud, and merchants are responsible for card-not-present (CNP) fraud. The merchants’ share of card fraud liability has been around 35% of all losses, Aite Group found. Since the widespread adoption of EMV chip cards, merchant liabilities have risen to 60% of card fraud losses.
Additionally, card brands monitor card fraud losses and total chargeback-to-sales ratios. And they’ll place merchants with high or excessive chargeback rates into programs until they can reduce their chargebacks. But these programs come with hefty fines and long timelines. Knowing this, Getty Images addressed the situation.
In 2010, a Getty Images consultant recommended Kount’s eCommerce fraud solution. Kount’s solution combines AI fraud prevention that uses supervised and unsupervised machine learning and a flexible policy engine to decision authorizations in real time. Supervised machine learning analyzes past transactions to predict the likelihood of fraud. Unsupervised machine learning analyzes anomalies to detect emerging fraud. Kount’s AI is linked by the Identity Trust Global NetworkTM, which has visibility into 32 billion annual interactions, to assess the trust levels for all payments, accounts, and logins.
Implementing Kount’s AI and machine learning technology was nearly effortless for Getty Images. Kount analyzed Getty Images’ chargeback data to learn the source of their fraud and identify patterns. Once implemented, Kount’s AI can produce Omniscore, which assigns a letter grade to every transaction. Fraud analysts can use the score to assess the likelihood of fraud in a given transaction. Getty Images, for example, used scores with other transaction attributes to create fraud strategies that provide effective defense and reduce customer friction.
When decisioning a transaction, Kount supports three primary outcomes: approve, review, or decline. Customers can configure the solution to meet their needs and requirements. Some merchants may only want to approve and decline. Others may want to deploy all three. Getty Images wanted the flexibility to control its decision-making process to deliver smooth experiences to good customers and shut down fraudulent accounts quickly.
An important component of any fraud system is its ability to flag fraudulent transactions. Once flagged, machine learning can learn from the knowledge and more easily detect fraud in the future. Kount’s flexible system supports both manual and automated uploading of chargeback data into its system, which uses supervised and unsupervised machine learning to inform fraud strategies. The combination of the two in Kount’s AI provides highly accurate risk scores in milliseconds.
Since Getty Images deployed Kount’s AI and machine learning technology, it has identified more fraud and seen fewer false positives. Using Kount meant that Getty Images could decrease fraud losses without declining good customer transactions.
With the implementation of Kount, Getty Images reached a total chargeback-to-sale rate of less than .01%. This figure represents a reduction of more than 90%, compared to Getty Images’ historical highest chargeback levels before Kount. Today, Getty Images averages a fraud decline rate of less than 0.5% of all authorizations with a false decline rate of 0.06%. On average, for every 2,100 authorizations, Getty Images declines 8.4 due to a high certainty of fraud. Of that, 7.4 are fraudulent, and only one is valid. And with Kount’s ease of use and reliable customer support, Getty Images has scaled its fraud prevention process with a small team.