What's in this article?
When you have a new product, a new service, or a new website redesign, success is crucial. There are many big companies who’ve run into website development problems in the moment of truth. Whether your business relies on an agile WordPress design or a customized enterprise solution, it’s worth learning from the mistakes of others.
Here’s our list of the 7 biggest website failures and how to avoid them.
7. Yahoo Mail (December 2012, October 2013)
Yahoo Mail was once the third-largest email service in the world. However, a series of fateful redesigns didn’t sit well with users and, with a host of new mobile competitors, the webmail client eventually dipped to ninth place by December 2017. So what happened with this redesign?
One analysis puts it down to two issues. First, there wasn’t really anything wrong with the previous design. Corporate leadership roles reportedly changed hands, and the project was initiated soon after. Users complained about both a late-2012 redesign and another attempt less than a year later. Tellingly, only a quarter of Yahoo’s staff wanted to use the new design.
Another factor was that this unwanted and unliked redesign moved the website closer to the look and functionality of competitor Google Mail. Yahoo users wanted Yahoo Mail, but they got something more akin to its competitor.
6. Pokemon Go (July 2016)
In the Summer of 2016, a Pokemon fever broke out over the new Pokemon Go game. But the new website ran into problems from a spike in user demand and then failed to recover quickly. The problem slowed the momentum of the new product, just when it needed it most.
According to BlazeMeter, Pokemon Go’s problems likely could have been avoided with better testing and preparation. While it’s hard to know how popular any new product launch will be, it’s easy to know the load capacity limit of your website with stress and soak testing. Outages happen, but Pokemon Go could’ve been better prepared with backup servers and locations. Additionally, the company could have tracked end-user performance and activity to better anticipate future network demands.
5. Canadian Immigration Portal (November 2016)
There’s also the matter of knowing a site’s continuous delivery performance. That’s because spikes in traffic can sometimes come out of nowhere, even when you’re not launching a new product, service, or website redesign. Such was the case of the Canadian immigration website portal in the wake of the 2016 U.S. election.
Disappointed voters flooded the website, and it crashed repeatedly due to a sudden and sporadic high interest. The spike in web searches was also reported by Google. While many people were probably searching emigration procedures in jest, the crashes were real. One technical analysis suggested learning about network weak points and setting up alert monitoring dashboards to stay on top of it.
4. Apple App Store (December 2015)
It sometimes seems like celebrity Kim Kardashian is always in the headlines. While an infamous photo shoot did not manage to “break the internet,” Kardashian did crash the Apple App Store in late 2015 with the release of her “Kimoji” app. The app was a set of Kim-themed emojiis that cost $1.99 in the App Store.
Forbes reported that at one point on the release date fans were downloading the app at a rate of 9,000 per millisecond. The crash affected not only downloads of Kardashians product but other users of Apple’s App Store as well. It’s likely Apple knew there would be high demand, as Kardashian’s mobile App game from 2014 brought in $43 million in sales in the first three months. However, it looks like they weren’t prepared.
3. Macy’s (November 2016)
If there’s one day per year a major e-commerce retailer doesn’t want their website to fail, it’s got to be Black Friday. Unfortunately for Macy’s, that’s exactly what happened in 2016, when both the brand’s website and mobile app crashed under heavy loads.
The fallout from the website crash was not insignificant. Shoppers who couldn’t shop and complete purchases during limited sales complained loudly across social media. So the brand lost sales as well as brand satisfaction. Again, load and performance testing could have helped prepare for the high traffic of a sales holiday.
2. Apple Store (June 2010)
The iPhone 4 seems like a dinosaur of a smartphone today, but it was cutting edge technology in June of 2010. And everyone wanted one. All at the same time. However the whole pre-ordering system — from the Apple Store itself, to a new phone-based order-taker software, to AT&T’s website, to Apple’s 1-800 number — ran into major technical problems.
The system seemed to get most caught up between Apple and AT&T’s system integration. AT&T was the only authorized U.S. carrier at the time (this is before T-Mobile and Verizon offered iPhone products), and buyers had to check their eligibility to buy a new phone on AT&T before purchasing through Apple. The phone system began rejecting calls, while Apple Store and the mobile phone-ordering software froze on a processing screen. Again the problem was a load and stress testing problem, which might have been avoidable with better preparation.
1. Healthcare.gov (October 2013)
Finally, probably the biggest website failure of the last decade is the case of Healthcare.gov. This was an important website launch that needed to work so that millions of people would have access to health insurance. Unfortunately, the site had a huge number of problems, damaging the public’s faith in government services and causing lots of frustration.
The problems on this website failure could fill a couple of blogs, but a few issues stand out. The front-end work (what users see) and the back-end work (what developers create to make the site manage an insurance sign-up) were not properly coordinated to work together. There was also a great deal of red tape that got in the way of letting the website databases do their thing. Eventually, the issues were fixed and the site now works, but the political cost was not insignificant.
Bonus: How BuzzFeed Didn’t Crash (February 2015)
But I will leave you on a happier note. It is not always true that a big spike in traffic means a downed website. Pop culture news website BuzzFeed provides such an example. In February 2015, the BuzzFeed website published several articles on an optical illusion where a white-and-gold dress looked like a blue-and-black dress. And who else weighed in but Kim Kardashian.
Interest in this topic spiked, and BuzzFeed handled a peak of 670,000 concurrent visitors on the site. BuzzFeed published a post on their response, explaining how they handled 41 million views in less than 24 hours. The system admin crew joked at one point about “throwing servers into the internet.” (Note: Our sys-admins ask that you do not actually throw servers as that doesn’t help.) The article explains how their team had added 40% additional capacity when needed:
We have heavily invested in infrastructure provisioning and scaling. We know exactly how to scale fast from running drills. We have seasoned sys admins who have gone through this before.
So in short, it pays to plan ahead and be prepared for the unexpected. While it may not be crucial that articles about a dress color handle a traffic spike, the example is applicable for all those businesses where website failures are more make-or-break. If you’ve got a new product launch, a new service launch, or a new website redesign launch, you don’t want to encounter a problem. Failing to plan ahead could be costly.
See how Kaleidico can help you build the website you want, at the price you want. Give us a call at 313-338-9515 or email firstname.lastname@example.org to learn how our full-service digital agency can help with all your website needs.