Critical.
Authoritative.
Strategic.
Subscribe to CIO Magazine »

Inside American Eagle Outfitter's 8-Day website nightmare

What happens when a business suffers an 'unprecedented outage'

HANG IN THERE. That was the first plea American Eagle Outfitters offered its nearly 1.5 million Facebook friends three days after its trio of e-commerce sites had crashed.

The online retail outlets - ae.com, aerie.com and 77kids.com, which target teens and kids-s uffered what American Eagle Outfitters (AEO) termed a "major incident" on July 19. "This has caused all of our sites to go down," a Facebook update said on July 22, at 1:26 p.m. "We're working together 24/7 to get them up and running as soon as possible. Hang in there....we'll make it up to you. Check back soon."

Those back-to-school shoppers who checked back too soon at the sites received this message: "We're making updates to our sites." This was no planned maintenance, however. Behind the scenes, AEO and its external hosting provider IBM were scrambling for several days as the sites stayed dark.

The reasons for the four-day outage (and subsequent four more days of technical aftershocks) related to backup and disaster-recovery technologies and processes that failed, according to several articles on retail tech site StorefrontBacktalk.com.

As the articles describe, atypical and concurrent failures with IBM's hosting servers and backup plans as well as with Oracle's Data Guard utility program ultimately proved to be the sources of problems. (Stephanie Oschwald, a spokesperson for Pittsburgh-based AEO, did not return several messages seeking comment. IBM and Oracle declined to comment for this article.)

[ Read about e-tailer Zappos.com's $1.6 million pricing mistake ]

On July 23, at 5:39 p.m., AEO announced on Facebook that the sites were A-OK. "We had a few glitches but we're back! We missed you while we were gone. Free shipping now thru Monday, July 26 11:59 PST. Automatic on your entire purchase (US and Canada only)."

However, noted StorefrontBacktalk.com, four more days of technological healing lie in wait for the AEO site, "when it limped along with various parts not functioning until Tuesday afternoon, July 27th."

An Unprecedented Outage

The shock of a 100-hour e-commerce outage now over and the damage done to their critical back-to-school sales period, executives of the $3 billion AEO have to be left scratching their heads in frustration--and wondering just how much "teeth" they had written into their service-level agreements (SLAs) with IBM. (For the record, StorefrontBacktalk.com reported that AEO had recently filled the position of "Manager--Business Continuity & Disaster Recovery," and not a moment too soon.)

AEO's websites weren't the first to go down unplanned, nor will they be the last. Dating back to the mid-1990s, history is littered with tales of site outages, complex IT fixes and lost revenue--embarrassing for the companies and career killers for those in charge. Any Amazon.com unplanned downtime, for instance, is still reported with as much zeal as a Jennifer Aniston "new boyfriend" sighting.

But the duration of AEO's site outages could be considered unprecedented for an e-retailing brand.

"In my career, I can't remember a mission-critical application that was down for four days," says Paula Rosenblum, an analyst at RSR Research and former retail CIO. "It's almost impossible to conceive--and I say that not as an analyst but as a former CIO."

Deals, Discounts and Disappointment

The timing couldn't have been worse for AEO, at the outset of its "back to school" summer push. In its stores, AEO's July performance was a "disappointment," according to a Zacks Investment Research report on retailers. The retailer had also lowered its quarterly earnings outlook in July.

The company not only lost online revenue during the outage, but also attempted to make up for the downtime by offering its customers deep promotions--such as this deal promoted on its Facebook page on July 30: "Thanks for hanging in there while our site was down...this Friday & Saturday (7/30 & 7/31) online only take 25% off your entire order & we'll throw in free shipping, too!... Thanks again for hanging in there!"

Brian Walker, a Forrester Research principal analyst who follows e-commerce, doesn't term the AEO outages unprecedented, but unusual. "Sites routinely will become slow and the customer experience and conversion rate may be affected," he says via e-mail, "but to be down for four days has a significant business and customer impact."

AEO's "direct businesses" (those three websites) rung up more than $340 million in sales in fiscal 2009, which accounted for approximately 12 percent of the retailer's overall sales.

Outsourcing E-Commerce Ops: Deal or No Deal?

Outsourcing e-commerce operations in retail is not a new proposition--and IBM and Oracle, for that matter, are not fly-by-night operations. "Using managed services or SaaS solutions for all or key parts of the e-commerce platform is routine today," says Forrester's Walker. "On average, eight hosts will be involved in supporting a typical e-commerce transaction."

Even with the site outages, the outsourcing proposition is still sound, Walker says. "There is no guarantee that doing this in-house is any better," he opines, "and in many ways can be much harder to do as resources, budgets and skillsets may be harder to have at the right levels."

However, it's important that a company outsourcing its operations not abdicate every decision and responsibility over to the hosting company. "Often in these situations there will be some shared responsibility between the business and IT-support staffs," Walker says. Trouble happens, he says, when something as simple as a failure to communicate an upcoming big promotion or event that will likely result in a spike in online traffic big enough to take down a website.

Interestingly, during the sites' outages, AEO's mobile site was still functional, according to StorefrontBacktalk.com. RSR Research's Rosenblum says that's a "good news / bad news" situation. The good news: The mobile site was up, and AEO could have done a customer redirect from the online site that was down to the mobile site, if they chose. The bad news: It would appear that AEO is running separate, siloed systems (for customer databases and queries, for instance) for what are, essentially, similar e-commerce channels, Rosenblum says.

"Our research shows us that too many retailers are in a similar situation," she says. "They spend an inordinate amount of scarce resources--IT and human--forcing multiple channels into synch. The opportunity costs associated with these activities are incalculable."

Forrester's Walker advises other retailers to use AEO's outage as a wake-up call. "Get your responsibilities, processes and systems together now, test them, and correct as needed," he says. "Having a clear plan in and of itself will pay huge dividends in the event something happens, and with the technology and architectures we have available today this can be done at a reasonable cost."

"Consider the downside of having the business offline for a few days during the holidays," he adds, "and the ROI is pretty clear."

Do you Tweet? Follow me on Twitter @twailgum. Follow everything from CIO.com on Twitter @CIOonline.

Tags e-commerceUtilitiesbackupAmerican Eagle Outfitterssoftwareinternet

More about Amazon.comAmazon Web ServicesAmerican Eagle OutfittersFacebookForrester ResearchIBM AustraliaIBM AustraliaOracleSalesforce.comYahoo

Join the CIO Australia group on LinkedIn. The group is open to CIOs, IT Directors, COOs, CTOs and senior IT managers.

Comments

Comments are now closed

Computerworld
ARN
Techworld
CMO