The Payment Management Blog

Manage Order Rejection Rates to Optimize Profits

After completing automated and manual screening of your orders, they either move into the fulfillment process or are declined.  Our 11th Annual Fraud Report tells us that over the past two years, many merchants have reduced the number of orders they decline due to suspicion of fraud with no overall increase in the average fraud loss rate.  Declining valid orders from good customers is sometimes referred to as “customer insults” and is also known as “false positives,” where an initial determination of high fraud risk turned out to be false.   Since many customers will either try to re-submit their order or place it with a different merchant, it is difficult to know how many false positives / customer insults your fraud process is generating – or how many customers you are losing.

Depending on what goods are being sold and the costs of goods sold as compared to their gross margins, different levels of order rejection are appropriate for different merchants.  High-risk goods with high costs and low gross margins typically have higher order rejection rates.  This can be attributed to higher fraud attempts for these types of goods as well as lower merchant tolerance for risk because  more additional “good” sales are needed to make up the loss on a single fraudulent order.  Alternatively, merchants with relatively high gross margins, a high cost of customer acquisition, or larger lifetime customer value tend to employ the lowest rejection rates.

We believe it is important to set targets based on risk tolerance and potential fraud cost, and manage the trade-off between rejecting valid orders and controlling fraud losses.  For merchants selling a wide range of goods with different costs and margins, we recommend setting order rejection thresholds, or other screening criteria such as velocity of purchase attempt for the top selling products that comprise the bulk of online revenues or losses due to fraud.

Department of Commerce Estimated Strong E-commerce in Q4 2009

Recent up-and-down activities on the global stock markets may not be indicative enough of economic recovery, but what about e-commerce trends? Following Q3 2009's numbers, here’re some latest statistics from the Department of Commerce. The Census Bureau has published the Q4 2009 quarterly report for online sales by U.S. Retailers, representing up to half of U.S. e-commerce. The rest are online sales that include services such as travel, and wholesalers and manufacturers selling direct to consumers.

The preliminary estimates show a strong re-bound in online sales by U.S. retailers in the last quarter of 2009.  Year-on-year growth was 14.4%.  This growth rate is similar to levels last seen in Q4 2007 prior to the start of the recession.   While the higher growth rate can be partially attributed to the comparison with a weak 4Q 2008, the last quarter of 2009 did set a new online sales record, estimated at $35.85 billion and comprised 3.8% of total U.S. retail sales.  The strong Q4 performance turned a flat to negative growth year for 2009 online retail sales into a positive 1.7% growth.

While the US economic engine is far from full recovery, this is one positive sign.

Review How You Review Part 3: Feedback on Team Performance

Now that you’ve looked into the productivity of your review team and the effectiveness of the tools they use, you might now look into the data that shows your reviewers’ compliance with the established review processes, including final outcomes of transactions. Merchants we work with that are improving review cycle time and process accuracy have a well-documented review process. This process becomes the baseline that they use to compare statistical samples of actual reviews for quality assurance purposes.  As you can imagine, establishing a comprehensive quality assurance program provides an indication of training and development issues, as well as supporting continuous improvement efforts. We also suggest evaluating the chargebacks received against transactions that underwent review. Although this process seems fairly logical in nature, our 11^th Annual Fraud Report indicates that 46% of merchants lack the mechanism or process to regularly evaluate whether a reviewer’s transactions resulted in chargebacks.

With the results of the analysis, you can get a good start on the new year by incorporating the lessons you learned over the holidays, and, if you haven’t already done so, build in some of the data collection and evaluation processes outlined here. With additional data, you’ll be armed to make better decisions, emphasize effective tool usage, and improve your reviewer behaviors throughout the year. More importantly, you will be ready for the next holiday shopping season . . . . 11 months and counting down.

Review How You Review Part 2: Measure Productivity and Effectiveness

Previously, I suggested that now is a good time to assess how to improve your manual review operations. Today we explore what metrics we should look into.

As a productivity measure, many organizations look at the orders reviewed per hour (or per shift). Typical reviewer productivity varies by size of the online business.  For large organizations reviewing 20 orders per hour per reviewer is common (see the CyberSource 11^th Annual Fraud Report for reviewer productivity for smaller online businesses).  One additional measure that has proven to be useful is the average time in queue, which looks at the difference between the time an order was decisioned and the time it passed from the automated screen into the review queue. If you’re able to determine the time that each of these transaction status changes occurred, you’ll be able to directly relate your reviewers’ productivity to customer experience commitments and objectives.

The other piece is to measure the effectiveness of your review tools. It depends on the number of tools you have in use, and the complexity of your review processes. If you haven’t already, you should establish a mechanism for tracking the use of your review tools, preferably with relationship to the transaction and its outcome. If you’re monitoring at the transaction level, remember that you should evaluate the effect of a particular tool, such as an external address validation service, in light of both the accepted good transactions as well as the fraud chargebacks. You should also identify whether reviews that were completed without the use of services that cannot be automated (for example, bank verifications, or customer contact) might be automatically screened and dispositioned using your existing automated fraud screening tools.

In the next post, I’ll discuss looking at feedback on the quality of reviews that were performed.

Review How You Review Part 1: The Time Is Now

With the holiday shopping season now behind us, your attention may be turning to the optimization of your manual review processes. You may want to find out, “how do I do things better or faster?” or “how can I improve the accuracy of my overall process?” These questions make great sense—you’ve probably just been through some of the highest volumes you’ll see for the year. You have full knowledge of what was done successfully – or not - in your operations, and you have accumulated valuable operational metrics for analysis.  Now is the time to sift through these details to pinpoint what in your process requires attention. In my next couple of posts, I will examine a few of the metrics you should look at, and how to parse the information.  Generally, these metrics relate to the productivity of your review team, the effectiveness of the tools and processes used in reviewing orders, and feedback on the quality of the reviews performed. Our manual review teams and their tools stand to gain the most in improvement when we analyze these aspects of our operations.