- Author By Miva |
- Posted on
This blog post comes from our partner, Yottaa. Yottaa provides end-to-end acceleration, protection and visibility for modern web applications across the wired and wireless web, delivering speed, scale, security and actionable insight.
Well, because Yottaa’s web monitoring service monitors tens of thousands of web sites and has accumulated a database of tens of millions of web performance samples, we have a treasure trove of data. We set out to answer the following questions:
- Are there ones that have a particularly poor “service level agreement” (SLA)?
First: As an Online Retailer, Does This Really Matter?
Web performance is often something businesses and website owners take for granted; many times, we assume that our site will run smoothly and efficiently, and when it doesn’t it can be incredibly frustrating.
And the truth is, if your website takes longer than 3 seconds to load, it’s already losing revenue. Every additional second after that represents a 7% loss in conversions, and after 3 seconds 40% of visitors will leave your site.
Now, let’s look at the numbers. With data across millions of samples, we looked specifically at the following metrics:
- Average size of the download;
- Average Time to Last Byte (how long it takes to download);
- If a widget is popular but takes little time to download, it may not have a particularly adverse impact on performance
- If the widget takes a long time to download, but is not particularly popular, few sites and visitors are affected
- But, if a widget that has a high percentage of slow download times, AND is popular, has a higher likelihood of screwing up performance
Here are some of the results:
- Google Analytics
- Google AdSense
- Google AdWords
Google Analytics: Tracking Global Speed
Should You Host on Third Party Domain or Your Own?
This data suggests that though there’s some variation, there’s not an obvious benefit to hosting it elsewhere, on average. (Let’s emphasize “on average” because there are approaches where third party hosting is beneficial – see next section!)
Knowing the size of the file, server IP address and the download timeline, we can compute the individual server’s bandwidth. And as it turns out, this varies widely – typically a couple hundred kilobytes per second, though (as illustrated in the histogram below) there’s a spike at the high end.
Three Tips on What to Do Now
- Consider leveraging a content delivery network