Amazon Not Primed to Handle its Own Shopping Holiday

Another Amazon Prime Day is in the books, and once again the #PrimeDayFail hashtag was trending on social media throughout the one-day cybershopping extravaganza. Where last year the complaints centered largely on lackluster deals, 2016 delivered a different frustration: checkout failures. It turns out Amazon.com’s technology team couldn’t cover the check its marketing department went to cash.

Failed Checkouts Make Buyers Check Out
With seemingly nonstop hype and promises of considerably more deals than the inaugural Prime Day, Amazon spawned a tsunami of demand, so much so that its e-commerce applications buckled as a result of the sheer volume of cart activity generated by eager buyers. When shoppers went to check out, they were greeted with the error message "Add to cart failed … Retrying," a screenshot of which was retweeted by numerous Twitter users who happened to double as irritated Amazon Prime members that day.

Shaky Foundation Upsets the Shopping Cart
What must frustrate the Seattle-based retail goliath is that it does so many things right to get itself into position to process millions of concurrent purchases. Already versed in driving up demand with competitive pricing on normal days, Amazon upped the ante by guaranteeing more compelling bargains than last year's Prime Day. Moreover, the company’s fulfillment processes, which were mastered long ago, were ready for the challenge. So what failed the online shopping leader? Its weakest link appeared to be its reliance on an older IT infrastructure — outdated technologies that require months worth of planning in order to execute at scale when the need arises. Even with its best advance preparation, Amazon was likely doomed to underprovision its database capacity, leaving it unable to process the even-higher-than-expected number of cart transactions Prime Day ultimately drew.

Modern-Day Databases ‘Flex’ to Meet Demand
So what could Amazon have done differently to avoid this second anniversary snafu? It could have taken advantage of newer database technologies that allow you to quickly increase capacity on short notice to handle the temporary spike in cart and checkout transactions, then scale back down when the extra “processing power” is no longer necessary (thereby avoiding wasted capacity and cost). Through its own internal forecasting efforts and ongoing social media monitoring, it’s likely that Amazon had a decent handle on expected demand the week of the event, but it was too late for the company to adjust thanks to its outmoded underpinning database. A modern IT database would have allowed Amazon to add server instances even as late as the day before Prime Day so that it could deliver a seamless shopping experience devoid of "Add to cart failed" error messages — and avoid the resultant PR/social media firestorm (again).

A Lesson for All Retailers
If Amazon, a relatively advanced company technologically, can struggle to handle Cyber Monday-levels of shopping activity, it serves as quite the cautionary tale for competing retailers — Wal-Mart, Target, Petco, Best Buy, Neiman Marcus, to name a few, some of which have had similar stumbles in the past. Amazon and these other large merchants have to figure out how to handle short-term cart transaction deluges of this magnitude if they want to stage (or compete with) Prime Day and/or similar large-scale online events, such as the release of the new “it” video game Pokemon Go.

Amazon will undoubtedly try to bring Christmas in July again next year, packed with can’t-miss prices farther than the eye can see (or the search browser can navigate). However, without a modern IT foundation, the bed it makes likely won’t be sturdy enough to lie in.

Mike Azevedo is CEO at Clustrix, a provider of the first scale-out relational database designed to meet the needs of large, fast-growing e-commerce sites and other big web- and cloud-based applications.