Web Performance Review of $5.3 Billion Weekend – LoadStorm


Tis the season….  Why haven’t large corporations learned the value of load testing yet?  How could e-commerce sites not be ready to handle the increased traffic for the busiest selling period of the year?  Unfortunately, with the high traffic of Cyber Monday, certain websites were unable to meet customer expectations.  Ok, let’s be honest – bad web performance cost some sites money and really made buyers angry.

Let’s conduct a review to see what we can learn.  First, a few relevant statistics:

  • For the five-day period from Thanksgiving through Cyber Monday, online buying from desktop computers totaled $5.3 billion, up 22 percent versus last year (source: comScore)
  • Smartphones and tablets drove nearly a third of traffic — and for some retailers, more than half (source: USA Today)
  • Sales on Amazon were up 44.3% over last year, while sales at eBay were up 32.1%, according to ChannelAdvisor
  • Fedex expected Cyber Monday 2013 to be their busiest day ever – shipping 22 million packages

Brand Damage from Poor Performance – Motorola

Motorola was probably the biggest loser in web performance this week.  Their discounted Moto X created high demand, and it drove a traffic spike in their website that caused failure. Apparently not many people were able to take advantage of the buying opportunity.  The CEO of Motorola posted an apology on their blog – asking forgiveness for the site outage.  He admitted  that their load and performance testing was inadequate.

“Our pre-sale site testing was not sufficiently extensive. Testing failed to reveal weaknesses caused by large volumes of concurrent orders flowing through the MotoMaker customization engine.” – Dennis Woodside

The damage was evident on social media very quickly.  On Motorola’s Facebook wall, shopper Cameron Horst wrote “I have been up since 5:45AM trying to purchase this phone. Once the promo went live, and moto maker wouldn’t work, I removed my motomaker from my cart and tried to purchase the dev. Unfortunately i was too late. Now i am locked out of motomaker with nothing in my cart. If i am unable to purchase a phone after all of this i will be furious.”

Motorola replied back, stating “We’re currently restoring online services. Some may have site access, but Cyber Monday deal prices won’t begin until all systems are a go.”

Cameron commented “They already did begin, and now Dev Editions are sold out. These robotic responses don’t help. Its unbelievable how poorly executed this sale was and it has negatively tainted my view of motorola as a company.”

It seems clear to me that Motorola cut corners on load testing, and the monetary impact will be difficult to calculate fully.  Whenever a customer uses words like “furious”, marketing people are cringing because they understand how much damage will linger for a long time.

Lack of IT Investment Costs Royal Bank of Scotland Millions

The RBS is facing millions of pounds of compensation claims because their systems failed during heavy load on the busiest online shopping day of the year.  The failure left RBS customers unable to use credit and debit cards for purchases.  The CEO of RBS admitted that their lack of proper investment in systems, and certainly load testing or web performance testing, caused the failure.

He called the failure “unacceptable”, and  he said, “It will take time, but we are investing heavily in building IT systems our customers can rely on.”

Apparently, bank cards issued by RBS and its subsidiary NatWest stopped working during Cyber Monday around 6:30pm.  Their customers began complaining that card transactions were not completing.  Not only online transactions were affect, but people were stranded at gas pumps and in restaurants too. It was reported by a British newspaper that the bank’s website and smartphone app had also experienced problems.

From my perspective, the most curious part of the whole fiasco was that a RBS press release denies the problems were related to the much heavier than normal Cyber Monday volume.  What?!

The CEO will admit that their company has been negligent for decades in IT investment, but they want to make sure everyone knows it is purely coincidence that their failure is not a scalability issue.  Are you serious?  That’s a bit like the scene in Monty Python’s Holy Grail where the knight claims, “It’s just a scratch” while his severed arm is gushing blood.

Perhaps this is a good sign.  My reasoning is that global companies are now recognizing that poor system performance due to scalability results in more brand damage than simply a bug in the code.  I may be totally wrong, but I’m speculating that the media attention to website failure under load (e.g. Healthcare.gov, Target.com, etc.) has raised awareness of the value of load testing and highly scalable web architectures.

Belk’s Site Crashed – “I’ll spend my money elsewhere.”

Belk failed to offer a wide range of consumer goods over the past week. Their website repeatedly crashed on Thursday and Friday, leaving shoppers to leave feedback on Belk’s Facebook page.

“What a joke Belk’s has been this Black Friday. Why weren’t they prepared for this? Black Friday only happens once a year. I was able to get online yesterday….and complete my order….only to press submit and get an error. I wasn’t able to get back onto the website at all yesterday after that. I have finally been able to get back online today. Guess what? Now the sale items are not at the sales price. Ridiculous. You would have thought that they would have kept the same pricing until the end of the day tomorrow because of THEIR mistake. Pathetic. I’ll spend my money elsewhere,“ said shopper Edmond Edwards.

Ouch.  No commentary needed.

Belk apologized, “Unfortunately, the site could not handle the high volume of Black Friday shoppers during peak times today. We value your opinion of us and want to assure you that teams continue to work around the clock to provide you with a better online experience.”

Too late – they should have invested in load testing BEFORE the failure.

Multinational corporation Google also had their own difficulties upholding traffic for their Motorola product.

Wal-Mart Complaints for 1 Hour Guarantee

Even the retail giant Wal-Mart had trouble with their online orders regarding their 1 hour guarantee. How the 1 hour guarantee works is, if the item runs out of stock in the first hour, the customer receives a card with an access code to access their deal online. After making a purchase online, it was intended that the items purchased will be shipped to the shoppers local Walmart before Christmas for the same price as on Black Friday.

 The issue was that when some items ran out of stock, customers were unable to access the Walmart.com/1hourguarantee website and therefore were unable to claim their ‘promised’ item.

There were hundreds of complaints on social media about the site being down, a few customers were even commenting that the site was never functional and that the whole deal was a just a hoax to attract more shoppers.

Poor web performance reduces value of “deals”

The Obvious Conclusion: Load Testing is Good

Healthcare.gov raised the awareness of web performance in October.  I’m quite sure that its failure under load was the best advertisement for cloud load testing that LoadStorm experienced in our 5 years.  But these stories of performance problems just keep coming.  Companies are still making the mistake of ignoring adequate investment in load testing and performance optimization.

When it comes to online shopping sites, slow speeds cost money.  Crashed sites cause brand damage.

The financial and social impact of poor preparation will hurt your company’s reputation. Online shopping is growing more than 20% annually, and it is tragic when buyers experience personal pain from a poorly performing e-commerce application.  It’s unnecessary.  It’s bad business.

Be a web performance hero – talk to us about how we can help you integrate performance testing into your development cycle.  We will openly discuss your situation and make recommendations, regardless if you use our load testing tool or not.

Similar Posts