Three Suggestions for Avoiding Automation Aberration

Sears and Amazon.com both recently came under fire for allowing rings decorated with the Nazi swastika symbol to be posted for sale on their third-party seller sites. While both retailers quickly pulled the items down and issued public apologies, they received a large amount of negative publicity and surely damaged their brand image with many offended customers, perhaps to the point of permanently losing some of them.



By allowing third-party sellers to post items using automated systems with little or no apparent oversight, two of the biggest names in retail suffered embarrassment and brand damage. This is a perfect demonstration of how automation, which is a useful and necessary tool for performing retail activities at scale, can create difficulties without proper controls. Here are three suggestions for Sears, Amazon and any other retailer looking to avoid automating themselves right into trouble.



Getting Personal



Quite simply, automation works best when paired with human instinct and insight. Andrew McAfee, co-director of the Initiative on the Digital Economy at MIT Sloan School of Management and a world-renowned expert on artificial intelligence, recommends that people “race with the machines,” or collaborate with computers. The combination of machine logic and human intuition has been proven superior to either entity by itself.



By having human beings manually review and check third-party marketplace listings on a regular schedule, Sears and Amazon may have been able to find and remove these inappropriate items before they “went viral,” offending consumers and damaged their brands.



The Exception Proves the Rule



Manual review will help catch errors that slip through automated systems after they occur, but exception alerts help catch errors before they are processed. Any modern automation system features exception alerting, which allows end users to set parameters that trigger automatic notification of items needing review. Certainly in the case of automated third-party marketplace postings, keywords such as “swastika” and “Nazi,” and if the system is capable any offensive symbols or images, should be marked for exception alerts that allow them to be removed from the content stream before they enter public view.



Following the Patterns



Analyzing patterns in statistics like Web traffic and social media chatter is another after-the-fact safeguard that can help detect public reaction to automation-related mishaps before they attract widespread attention. On the way to causing public outrage that reached mainstream media outlets such as the Wall Street Journal, the swastika ring listings on Sears’ and Amazon’s marketplace sites attracted a lot of attention from horrified consumers who did not make a purchase, but did tell their friends to check them out and also posted negative comments on social sites like Facebook and Twitter.



By carefully tracking consumer activity on their marketplace sites, Sears and Amazon could have detected an unusually high number of short-duration visits to the offending ring listings that did not result in purchases. Also, there are now numerous “social listening” solutions and hosted services that help retailers detect pertinent trends in social media discussions.



Looking for unusual patterns in digital traffic and conversation pays many dividends for retailers, but staying ahead of negative response to automation errors (which may involve assortments, prices or other factors in addition to online product listings) is certainly one of them.