Follow Datanami:
November 20, 2017

Data Wrangling Enlisted to Fight Black Friday Fraud

Black Friday and Cyber Monday are coming, and the fraudsters are ready.

With a growing percentage of mobile applications and app stores listed as malicious in the run up to Black Friday (Nov. 24) and Cyber Monday (Nov. 27), retailers are looking for ways to keep up with new fraud detection tools. Among the bottlenecks to fraud detection, data preparation specialists note, are data wrangling tools designed to among other things speed the development of fraud models used to spot fraud and alert retailers.

With memories of the Target (NYSE:TGT) data breach of late 2013 still fresh, security firms and data prep specialists are highlighting the heightened holiday shopping season threat and pitching solutions. Digital threat manager RiskIQ, for example, reckons the top five online retailers have more than 32,000 “blacklisted” apps associated with their brands. The suspicious apps attempt to trick shoppers into entering credit card information or download malware that steals information.

As fraudsters become more sophisticated, retailers reliant of big online sales during the holidays are scrambling to keep pace. Data prep specialists such as San Francisco-based Trifacta are promoting data wrangling approaches designed to help developers accelerate the preparation of data used to constantly update fraud models.

Trifacta, which last week announced that customer Consensus Corp. has implemented its data wrangler, said Consensus is using the data prep tool to accelerate development of models used for fraud detection. Previously, the Target subsidiary relied on a team of developers to whip data into shape for fraud models. The problem was time lag between emerging threats and the weeks required to organize data used to write new scripts.

Fraud model development took on average about six weeks, delaying “the company’s response time to fraud threats with the most up-to-date information,” Trifacta explained in a blog post.

Even Trifacta has had problems keeping up with the volatile online retail sector. As the market for self-service data preparation explodes, Trifacta launched a data wrangler partner program last year to expand its reach. It also works with a growing list of companies that include the top three Hadoop distributors along with visualization tools providers and other data preparation specialists.

“As organizations scale their machine learning initiatives, we’re seeing data wrangling become a huge bottleneck to executing these initiatives efficiently and accurately,” Trifacta CEO Adam Wilson noted in a statement. “Preparing data requires scripting capabilities that impede [business intelligence] or product teams from guiding and accelerating this process.”

Those tools will be tested over the next several weeks as online retailers hustle to parse more threat data about fake apps and landing pages promoting deals, coupons and other scams. RiskIO, the threat management vendor, estimates that one of every 25 Black Friday apps is a fake.

The stakes are high: Market watchers estimate that consumers spent more than $9 billion during last year’s Black Friday shopping frenzy, of which $1.2 billion was generated by mobile shopping apps. Meanwhile, fraud is costing retailers even more as expensive devices are sold online. Trifacta estimates retailers must generate five additional transactions to cover the cost of a lost sale stemming from fraudulent account.

Recent items:

Trifacta Brings Partners Into Data Prep Fold

Survey: Data Analytics Falls Short in Detecting Fraud