Analyzing Black Friday and Cyber Monday Disasters

By Saoirse Hinksmon

Thanksgiving weekend has evolved into the biggest shopping days of the year, with Black Friday and Cyber Monday deals enticing people to purchase gifts early as well as items they’ve been coveting for months. This year, Black Friday online sales soared to $6.22 billion in the U.S., a 24 percent jump from a year ago and a record high, according to Adobe. Cyber Monday smashed last year’s spending record, with a 19 percent increase, coming in at $7.9 billion.

However, it wasn’t all gravy for a slew of retailers this past weekend, as downtimes and outages plagued a handful of companies. Outages included popular sites such as:

J.Crew;
Ulta;
Hulu;
Carter's;
Walmart;
Lowe's;
Macy’s;
Costco; and
Nordstrom.

Needless to say, people were not thrilled …

So What Happened?

I set up 40 or so monitors of my own to track website performance all over the world for companies in a variety of different verticals. I used the synthetic monitoring tool from SmartBear, AlertSite. Interestingly, across all the monitors, the most outages took place on Black Friday around 4 p.m.

What caused these sites to slow down? One culprit that was easy to identify right off the bat for performance slowdowns was third-party tags. Tags are small snippets of JavaScript code or small transparent pixels embedded on a page that can be used to collect data from a website. These typically include tags to help with user tracking, analytics, personalization, chat features, and more. While tags offer an immense amount of benefit like visibility into user actions, customization, and engagement, finding a balance for tags and website performance is essential. You can see in the screenshots below that on Macys.com and Bloomingdales.com, the top five slowest objects are third-party content.

Third-party content can have detrimental impacts on your overall page responsiveness. Check out what happened to Bloomingdale’s baseline:

Or Nordstrom’s performance deviation once it hit midnight after Cyber Monday:

Understanding the objects on your page, the impact they have, and optimizing around those is key, especially on big sale days like these, where you know your website traffic will differ dramatically in contrast to a regular shopping day.

How Can Retailers Improve?

Avoiding slowdowns and outages seems like a daunting and impossible task, even for the highly prepared. However, this doesn’t have to be the case. Testing and monitoring synchronously for both performance and functionality help ease the response when something does go wrong. For example, if you’re able to reuse test cases or scripts in production, identifying the point of failure, then reproducing the problem becomes quicker and keeps mean-time-to-resolution (MTTR) low, ensuring your visitors get the envisioned web experience as soon as possible after something goes wrong.

At a minimum, retailers should be load testing and monitoring their website performance. Understanding how your site behaves under load and alerting when something is amiss in production is the true key to success. There are performance tools on the market that are easy to try and cost efficient to buy, especially considering that one minute of downtime can cost you $5,600. So consider the overall return on investment, and map out a testing and monitoring strategy that empowers your team to optimize for both performance and MTTR.

Saoirse Hinksmon is a digital content marketing manager at SmartBear, an information technology company that delivers tools for application performance monitoring, software development, software testing and API management.