Trick or Treat: How did the Online Costume Retailers handle the Halloween traffic?
  October 31, 2013

Halloween costumes have become more intricate than ever before. According to a National Retail Federation survey, Americans are expected to spend $2.6 billion on Halloween costumes this year. We will spend $1 billion on children’s costumes, $1.2 billion on adult costumes and $330 million on pet costumes. Also, so that you’re not wearing the same costume as anyone else tonight, more than 5 million adults plan to dress as a witch, and 2.9 million as a Batman character. All these purchases have put quite the load on servers.

Unless your creative side kicked in and you’ve made a costume at home, you turned to a retailer to buy a costume. We’ve investigated how the demand for more Halloween costumes has impacted performance of the websites. To conduct our investigation, we monitored a few of the top sites that consumers were most likely to visit to buy a costume to collect the performance data from a 3-step interaction.

Did you shop for costumes online?

Below are the results for an online interaction that visits the home page, searches for a product, and views the product detail:


availability halloween post

Response time halloween post

Overall the performance of the sites was impressive. The response times ranged from 7s to 23s and the availability average was around 99.5%. All the sites handled the holiday traffic very well and no major incidents were encountered - but it was that took the lead with a 99.90% availability and a response time of 7.7141s. No wonder they are listed as one of the best Halloween costume stores online this year.

Now, let’s look at the data collected from the home pages form a visual user experience perspective:


Full Page Load Time (seconds)

First Paint (seconds)

Above the fold (seconds)























Taking only into consideration the full page load time of these sites, would indicate that they all fail to meet the user’s expectations page load time of 2 seconds or less – right? But as we analyze the user experience metrics, the First Paint time indicates that the sites began to render visually for the users in just over a second except for Amazon that kept it under a second.

Take a look at the Amazon example below:


Full page: 5.2343s

Halloween Post Availability Full Page

First Paint: 0.9024

Above the fold: 2.3984s


Fast page load times mean happier users

When consumers conduct online transactions, they expect pages to load instantly. They always want fast content delivery and if your site is not able to deliver you could lose 40% of your users. Whatever they are looking for on your site, they will go and look for it elsewhere - usually your competitor! Research shows that fast page load times mean happier users – site speed also impacts revenues, conversion rates and page views will be impacted in a positive way.  Site performance enhancement ought to be done on a regular basis.

Retailers should aim to increase conversion rates, decrease bounce, increase page views and time on site. I’d like to leave you with these good recommendations that could help achieve these goals:

  • Benchmarking: Quantify what is really going on with your site relative to the direct competition. Look beyond the home page. Pay attention to all of the interactions that lead up to conversion.

  • Load Testing: Prepare your website for demanding amounts of traffic by performing load testing to test and verify the scalability and response profile of your website before peak demand hits.

  • Third-party content management: Because you have little control over third-party content stick to best practices and load 3rd party content asynchronously.

Halloween is certainly not the biggest retail season of the year. Cyber Monday is just around the corner, and in case you haven’t begun, you should start preparing now. What other steps do you take to prepare your site for high-traffic times?

See Also:

[dfads params='groups=934&limit=1&orderby=random']

[dfads params='groups=937&limit=1&orderby=random']