Blog Post

Black Friday and Cyber Monday Performance 2019

Overall the performance of popular online retailers was better this year when compared to the last few years.

The November chill has settled in, holiday lights are up, and the Christmas hits are on loop again. This can only mean one thing: ’tis the season of deals and discounts! The holiday sale season started early this year with several retailers going live with Black Friday deals several days before Thanksgiving.

It is also the time of year when Catchpoint gets you up to date on everything that went down in the digital world, performance-wise, over the Thanksgiving weekend. We monitored over a thousand websites last weekend; in addition to helping our customers monitor applications, we also tracked the performance of popular retailers. In this blog post, we review some of the performance issues detected over the last week.

Although most of ecommerce and online retailers were prepared well in advance of Black Friday, others were still caught off guard when traffic volume spiked.

Website Performance for the Week

Capacity Issues

The extra load that holiday sales bring is one of the main causes of performance issues for online retailers. Servers are unable to handle the influx of traffic, which can cause a major bottleneck in the application delivery chain. Costco was among the first retailers to hit this performance roadblock.

In the graph below, the spikes in “Time to First Byte” (i.e. the time it took to receive the first byte of data from the primary URL) impacted page load time. This issue began Wednesday and continued throughout Cyber Monday.

Image: Line Charts Displaying Performance and Availability of Costco Website

While Costco’s homepage was available, the search results page, product details page, and the shopping cart page slowed down considerably.

Image: Screenshot of Costco Message to Customers

H&M experienced performance issues that started around 5 am EST on Thanksgiving Day. The scatterplot graph below highlights the slow performance on the product details page.

Image: Scatterplot Graph for H&M Product Details Page

The situation deteriorated for H&M who was soon unable to manage the high volume of incoming traffic. As a result, the retailer displayed a message to users that they were experiencing a temporary issue.

Image: H&M message displayed on Product Details Page

In addition, high wait times were detected for H&M’s search results page as well, which prompted the following message to visitors:

Image: H&M message displayed on Search Results Page

Interestingly, H&M experienced similar performance issues during Back Friday in 2017.

Image: Scatterplot Graph of H&M Performance from Black Friday 2017

Similar problems were found on Hobby Lobby’s website. On Sunday, December 1 beginning around 1 AM EST, Hobby Lobby’s search page went down and remained unavailable for over three hours.

Image: Scatterplot Graph of Hobby Lobby’s Search Page

The search page outage was caused by an internal server error and led to site visitors seeing HTTP 500 messages like the one below:

Image: Screenshot of Hobby Lobby Message About Performance Problems

Capacity issues contributed to performance problems for Costco, H&M, and Hobby Lobby throughout the holiday shopping weekend. Drilling into the data reveals the amount of downtime each retailer experienced because of capacity problems:

  • Costco: 17 hours
  • H&M: 10 hours
  • Hobby Lobby: 4.3 hours

In the graphs below, we can see drops in availability correlate to high response and webpage response times. The slow performance of mission-critical conversion points in user journeys – search page, product detail page, add to cart, etc. – highlight the need to monitor more than binary availability measures to determine whether or not your site is reachable.

Image: Line Charts Displaying Downtime Measurements for Costco, H&M, and Hobby Lobby

According to Business Insider, Costco may have lost as much as $11 million in sales due to downtime, though extending the length of sales promotion is likely to recoup a large portion of that figure as users return to complete their purchases.

Connection Issues

Connection issues proved to be another pain point for retailers. Forever21 was briefly impacted by high connect and wait times on Saturday, November 30. More specifically, Forever21’s website was unable to load images and scripts served from their CDN, which impacted end users for approximately 15 minutes. The waterfall graph below shows the high connect time experienced by the website:

Image: Waterfall Graph Showing High Connect Times on Forever21’s Website

Sephora experienced performance volatility from a specific host, a third-party community integration, on the page. The community platform used by Sephora experienced performance issues, which in turn slowed down the site.

Image: Scatterplot Graph of Sephora’s Homepage, Search, and Community Pages

On Thanksgiving Day, Home Depot’s website slowed down when issues arose while fetching content from a particular host mapped to Google Cloud. The scatterplot below identifies issues with a single host:

Image: Scatterplot Showing Slowdown Across Home Depot Website

In the waterfall chart below, we see a high wait time (in red) for each request, which impacted overall site performance.

Image: Waterfall Chart Showing High Wait Time (TTFB) on Home Depot Website

What Forever 21, Sephora, and Home Depot’s performance problems highlight the importance of proactively monitoring server connectivity across increasingly distributed digital infrastructures and networks.


Overall the performance of popular online retailers was better this year when compared to the last few years. We did not detect any high-impact or widespread third-party outages, and no websites were completely down during the busiest digital sale period of the year.

In the run-up to the holiday sales season, retailers added new content, scripts, and features to optimize website performance. But these new additions can easily break if not properly tested and configured. To ensure optimal end-user experience, it is vital to test the different features or modules in the application for functionality. A/B testing new features and monitoring the website by simulating real-world bandwidth conditions is a necessary step in preventing issues from impacting website performance and / or end-user experience.

As the holiday sale season swings into full gear, here are a few performance tips to remember:

  • Invest in scalable infrastructure to ensure bandwidth is available to handle the traffic spike

  • Monitor CDNs and third-party vendors

  • Test new features and modules well ahead of time. Make use of A/B testing.

  • Optimize page size and content.

  • Proactively monitor beyond uptime to maintain availability, reachability, reliability, and performance.

  • Adopt a monitoring strategy that provides visibility across every layer in your application delivery chain.

This is some text inside of a div block.

You might also like

Blog post

Windows 11: Run a better traceroute

Blog post

Prioritize Internet Performance Monitoring, urges EMA

Blog post

Traceroute InSession: A traceroute tool for modern networks