As published on PYMNTS.com Powering the Digital Shift 2020, June 30, 2020
In a 2019 PYMNTS eBook much like this, we made predictions for eCommerce in 2020 and the next decade. I recall writing about the role that artificial intelligence (AI) would play in protecting the complete customer journey.
Just six months into the new decade, we’ve seen that play out in ways we couldn’t have predicted in 2019. With the pandemic spurring overnight digital transformation and innovation, the roles that fraud prevention, customer experience and AI play in commerce have grown faster than ever. This introduces new opportunities, as well as reasons to reject the status quo and lasting lessons.
Opportunities: New Channels of eCommerce
With an unprecedented number of physical locations closed, brick and-mortar retail and restaurants needed to adapt quickly in order to stay open. For many, this meant expanding new channels of eCommerce, such as buy online, pick up in-store (BOPIS), delivery offerings, mobile order-ahead and more.
Some of these companies experienced a steep learning curve — not only did they need to find a way for customers to shop and pay, but they also needed to create a method for pickup, delivery and authentication. During this shift in behavior, a negative experience could cause a loyal customer to abandon a brand or switch to a more convenient option.
Rejecting the Status Quo: Cumbersome Manual Reviews and Customer Insults
In uncertain times, customers want their necessities to arrive quickly. For example, expedited shipping increased 305 percent year over year in mid-March as people began to stock up. That urgency doesn’t allow for cumbersome approval processes. Fast, accurate approvals are also essential for mobile order ahead at restaurants, as consumers often expect meals within minutes of placing their orders.
Worse than a long manual review is the risk of insulting a good customer by turning them away because of a false positive. Unsophisticated fraud prevention strategies couldn’t adapt as the pandemic changed both how people shopped and what they bought. Suddenly, ordering 10 bottles of hand sanitizers at 3 a.m. might not be as unusual as it once seemed. Supervised machine learning or Brad Wiskirchen rules engines alone can’t catch these nuances in real time. Next-generation AI that combines both supervised and unsupervised machine learning with a robust data network of trust and risk signals can better identify what good or risky orders look like today.
This means eCommerce businesses, whether well-established or just entering the digital learning curve, must ….