By: Justin McDonald
Fraud and risk management strategies tend to focus so much on automated risk decisioning that improving manual review performance is often an afterthought. Consider the cost savings and increased revenue an organization could realize by cutting average order review times while also reducing sales insults and missed fraud on reviewed orders. This is why improving performance of manual reviews is at least equally important as efforts to reduce order review rates.
Manual reviews are a critical piece of most all eCommerce fraud and risk management strategies, but also often a challenge. They are operationally intensive while performance metrics typically differ quite a bit across review agents, over time and due to additional hiring during seasonal peaks. Organizations tend to focus the vast majority of their risk management strategy efforts on automated risk screening, but improving performance of manual order reviews is very likely to be a source of significant potential uplift.
The order attempt manual review rate is a critical key performance indicator (KPI) that most organizations measure and strive to maintain or decrease. Merchants are much less likely to consider, however, the magnitude of operational cost savings and increase in sales conversion if they could improve their average order review time and sales insults on reviewed orders by 10 or 20 percent each.
Maintaining a manageable order review rate is important, but often comes at a cost. Effective risk management is about striking a balance between fraud prevention and enabling sales conversion. All else equal, reducing order review rates comes with a sacrifice: either increased fraud losses or increased sales insults.
The often overlooked approach is to improve manual review performance. Providing order review agents tools and features that will expedite the order review process and decrease average order review time can be equally or more meaningful operationally than reducing the number of orders queued for review. Similarly, improving the accuracy of manual reviews, such that fewer reviewed orders end up as missed fraud or sales insults, can provide as much or more financial uplift than seeking to reduce fraud losses via automated screening.
Five tips for increasing performance of manual reviews are discussed in this article, which includes both things organizations could do in-house and things that would require the use of a third party provider. These may be tools or services that replace incumbent providers or ones that are additive to an existing risk management strategy, but in either case, an increase in financial cost would be justified by improved manual review performance. For the vast majority of organizations, labor is the largest cost component of performing manual reviews. If these risk management strategy additions, in-house or third party services, can increase operational efficiency as well as reduce sales insults and fraud losses on reviewed orders, they will be well worth the investment.
1) Limit the number of platforms or interfaces a review agent needs to use.
One of the major factors contributing to the average time it takes to complete a manual review is the number of different systems or platforms the review agent needs to bounce between to get all of the relevant order and identity data required to reach a confident decision on the order. Ideally, this is limited to one internal data system and one or a small number of third party service provider platforms.
There is great value in having one hub or system that contains all internal merchant order data and transaction history. It is common, however, for review agents to have to access one system for current order data, another system for historical data to perform order linking and transaction history analysis, and sometimes even secondary data repositories to access deeper order or identity data points. The more systems, tabs or windows a review agent needs to have open and perform a search or query from, the more cumbersome and time consuming each review becomes.
2) Provide recent and historical transaction link analysis and drill-down tools.
Often organizations maintain multiple order and customer databases, and one hub or interface that makes all of this data searchable and available in one place is an incredible asset. Link analysis and transaction history search or drill-down tools take this asset to the next level. Link analysis refers to the connection of the queried data point to other order attempts and other data points associated with each of the other order attempts, while the drill-down tool refers to the ability to quickly access this via a clickable icon or text link.
Each time a review agent needs to copy and paste or type something in to search, it takes a few seconds. Each data point is a potential clue; review agents determine which clues might tell them something meaningful, then dig deeper into that data point. Link analysis tools greatly reduce the time it takes to investigate each clue. Rather than copy-paste or type to search for transaction history on a phone number or shipping address, a great internal order data platform will provide clickable links. For example, when an agent wants to see what other order activity has been associated with a phone number, they just click on that phone number and a new window or pop-up appears.
Identity reverse lookup providers often provide these features within their web lookup portal or interface, where clicking one data point provides information on connected and related data points as opposed to information from the merchant’s transaction history. For example, clicking the shipping address may show a related address that is also known to be owned by the same person, but only the related address is associated with the phone number the customer provided. The review agent can quickly determine that this is a legitimate customer shipping something to their second home and promptly accept the order.
The impact on average order review time when adding these kinds of features and capabilities is obvious, but it will also impact review accuracy. When a review agent can perform deeper checks more quickly, they are more likely to investigate each potential data clue.
3) Provide visual-based tools and features that show connections and history of identity data points within the interface.
There are several identity authentication and reverse lookup providers on the market and each is differentiated, not just in terms of their data network but also with their user interface, features and tools that support performing manual reviews. A reverse lookup provider primarily designed for providing an API service will focus a lot of energy in the breadth and quality of their data network, but their review portal or interface may be lacking.
Visual cues go a long way in making manual reviews more efficient, and these come in many forms. This ranges from color coding (such as green and red) data points based on their risk profile or whether or not they match, to the use of icons (such as green check marks or a red X), to overlaying data points on a geographic map, to link or connection mapping. Each of these provides review agents visual cues that support their order outcome decision and guides them on how to proceed with conducting their manual review, improving both time and accuracy.
A geographic map can layer on the IP geolocation, shipping address and physical address associated with a provided phone number. This provides a visual reference point for a review agent to perform a triangulation analysis and consider the physical distance between these points. Identity link graphs or mapping can provide an engaging visual element, showing each data point a user presented and highlighting or connecting only those data points that actually have some association validated in the third party provider’s data network.
These are critical features bridging the gap between the insights provided by an identity solution provider’s data assets and how merchants actually leverage and best utilize these data assets.
We understand that manual review is about accepting good transactions with mixed risk signals. This is why our VERIFY product features a single screen that provides analysts clarity about all the identities, their connections and other trust & risk signals involved - in one short glance.
Matthew Hertz, CEO, Pipl
4) Provide consistent training specific to your organization and the review tools and services you provide review agents to use.
Thorough training and onboarding processes for new hires are critical in maintaining strong and consistent performance with conducting manual order reviews. Training should be at two levels: First, a baseline training that focuses on eCommerce fraud along with identity authentication and verification at a more general level. Second, training that is more specific to the merchant’s business including targeted products, risk profile and the specific tools and services utilized for performing manual reviews. The first level of training provides a foundation and understanding of eCommerce fraud, challenges with validating identities in the eCommerce channel and why it is important to detect and prevent fraud. The second level is more applied and should demonstrate how to use internal business systems and the vendor services and portals that review agents will actually be using.
Thorough training doesn’t replace the need for strong in-house systems and third party services, rather it provides understanding to make review agents better equipped to utilize them. There is an old idiom that states “a lousy carpenter blames their tools.” Training is critical in helping review agents make the most of their tools. Considering that there is typically a high degree of turnover among order review agents and it is a position that requires additional seasonal hires, having thorough and consistent training for all new and seasonal hires is critical .
5) Maintain KPIs to measure manual review health both overall and across individual agents.
Lastly, it is important to understand the health and performance of manual review operations, which requires data and reporting. Organizations should not only measure several metrics related to manual reviews and set internal benchmarks, but also look for fluctuations that show improving or deteriorating performance, which may initiate corrective action.
Important KPIs to track include: the share of order attempts queued for review, the share of orders queued that are reviewed (as opposed to purged), the approval and decline rates on reviewed orders, average order review time, and the missed fraud rate on reviewed and accepted orders.
Many of these metrics can be tracked on an individual review agent basis in addition to organization-wide. This can help identify which review agents may need more training, based on longer average review times or higher review-and-accept missed fraud rates, for example. Be cautious sharing these metrics with the team, however. Incentivizing a low review-and-accept missed fraud rate could lead to a higher review decline rate and increased sales insults.
Pipl's continually updated index of over 3 billion highly corroborated identities comprises more than 2.6 billion phone numbers and 1.5 billion email addresses, providing unmatched global coverage that includes 97% of U.S. adults. Pipl API and Pipl SEARCH allow e-commerce businesses to verify, investigate and resolve anomalous transactions via email or reverse phone lookup, so that our clients enjoy fewer chargebacks, a more predictable revenue stream, and happier customers.