Shades of George W. Bush and Victoria’s Secret all in one. We got fooled again last week by our own ineptitude and inability to learn from history when retailer Target’s website crashed under the weight of a highly successful marketing campaign. The last time anything remotely similar happened was when Victoria’s Secret decided to do an online fashion show.
The results and root causes were the same: too many eyeballs, too little bandwidth.
In case you didn’t see this, according to a Sept. 13 story in The New York Times, customers with ravenous appetites for the Italian designer Missoni’s clothing couldn’t wait to get at the bargains, and the Target site was no match. (I believe Oprah pronounces that “Tar-zhey,” whatever.)
This is not the first time Target has done this kind of promotion. According to the Times, Missoni is just the latest in a line of low-price designer self-knockoffs in which Target gets design houses to make some of their trademark clothing at Target prices. Previous houses that attempted to make it up on volume include Calypso St. Barth, Liberty of London, Rodarte and Zac Posen. As the character Cliff from the old “Cheers” show once said on Jeopardy, none of these people ever sat at my kitchen table, but that’s just me.
Get Off of My Cloud
What was different this time was the approach Target used. The retailer launched a national print and TV advertising campaign for Missoni, and the campaign was the first test of Target’s new strategy to take its website off the cloud and host it all in-house. (For this they get points for, uh, oh, never mind.)That’s right, this was cloud computing in reverse! In the middle of the Times piece was this hilarious paragraph:
While Target recently stopped using Amazon’s e-commerce platform to make its Web site work, analysts said Tuesday’s problems were probably caused more by unexpected demand for the Missoni products.
Which was immediately followed by this outright hallucination:
“It’s hard to attribute the site being down to the Amazon separation, although Amazon Web Services offers elastic computing power that would have helped with this surge in demand,” said Matthew Nemer, an analyst at Wells Fargo Securities who covers e-commerce, in an e-mail.
Unexpected demand? That’s the point! The demand was greater than what Target normally sees on the day after Thanksgiving, the busiest day of the shopping year. Of course it was unexpected; that’s why you use cloud computing — because you can’t forecast, and you presumably care, and because your job might be in the balance! Because you don’t get a second chance, and you really, really make your customers unhappy with an outage like this, and that’s not exactly what they teach in Retail 101. (Do I have any money with Wells Fargo Securities? No? Good!)
Of course this went viral and social in the worst ways. Memo to retailers who think social hasn’t caught up to retail — and I feel like curation software is really the better way handle this — here’s why sentiment analysis is so important.
Item No. 1: Jessica Alba to Jessica Simpson, also in the Times article,
“I dreamt about the Missoni 4 Target bike last night,” Ms. Alba posted, referring to a [US]$400 bike covered with zigzags.
“I want that bike too!!! So cute!” Ms. Simpson responded.
Item No. 2, again from the Times, actress Mindy Kaling posted on Twitter,
“@target broke my heart today when I could not access their site once in 9 hours for Missoni.”
(I am beginning to feel like David Letterman by counting these things. Also, I hate the idea of breaking these actresses’ hearts. So cute!)
Items Nos. 3 to infinity, a typical shopper posted,
“Seriously so mad at @TargetStyle. Why did you not prepare for the Missoni for Target line?
Why not indeed?
Yup, the Times ran that one too.
About the only really interesting and good thing for Internet commerce to come out of this was that in true social form, people were helping each other as much as they could. Some of the tweets suggested hitting the mobile site, which is apparently on a different server farm. Also, by the end of the day, Internet entrepreneurs were selling north of 6,000 Missoni items on eBay.
I’d like to thank Target for running this experiment to show us all the value of cloud computing, especially its elasticity. Thanks also for demonstrating the value of social networking and the importance of maintaining positive customer sentiment.
We’ve come a long way since we first started talking about “clicks” and “bricks” and how the Internet would be “the end of retailing.” Funny, but at the time I don’t think anyone envisioned the words being arranged in quite the way that happened last week.
At the end of the day, the Target "website down in 2011" is something that many online merchants struggle with and was unfortunately made worse by the timing during the re-launch. Although it is very difficult to avoid what happened, ultimately there are a few important considerations for merchants changing platforms that I’d like to share.
1. Licensed vs. Cloud. When making a decision to take the platform technology and delivery in-house (ie. going with a licensed software and hosting it), a merchant is now accountable for uptime, performance, capacity, and security of the systems. Most true Cloud, or Software-as-a-Service (SaaS) platforms provide the delivery technology along with the ecommerce software itself in order to specifically control the "quality of service" experienced by the merchant with their software. In addition to the "tech stuff" many mid-market and enterprise SaaS eCommerce Providers offer aggressive Service Level Agreements that hold them accountable for these key performance indicators. "SLAs with teeth" (as they’re often referred to) will incent the provider to keep the site performing in top condition.
2. When re-launching a website, large or small, it is important to understand that there is going to be a laundry list of small things that need to be address after the site goes live. Since it is impossible to fully replicate "real world traffic" (no matter how much money you throw at the project as with Target I’m sure the budget was significant for this project), and as a result it is impossible to fully understand what the real load on the system will be. So let this be a lesson to other merchants to use simple tactics to avoid crushing the website and annoying shoppers on the first day.
Lesson 1. Soft Launch (delay the national advertising campaign by a week or two so you can see how the system is performing)
Lesson 2. Phase the Launch (start with core functionality than phase in new categories, products, and features as you know the system can handle it)
Lesson 3. Monitor and React Quickly (build dashboards that allow top products and pages to quickly alert IT of load issues)
In summary, it is impossible to avoid bumps and bruises during the re-launch of an ecommerce website. As someone who re-launches 8-12 websites per year, I can say that these projects are significantly complex and the team is often under immense pressure from management and given unrealistic timelines. However, merchants of all shapes and sizes should understand the critical difference between "owning and operating" ecommerce software vs. the new age of "cloud/SaaS" ecommerce platform providers. And finally, by taking simple steps during a re-launch a merchant can reduce the risk of an outage and keep shoppers spending money.