Skip to Main Content

What Your Tag-Dependent Optimization Software is Hiding from You

Posted on 7.15.2013

By Jeffrey Vocell, Product Marketing Manager, SiteSpect

We’re going to be very transparent in this article about how JavaScript tags could be affecting your website and testing campaigns in ways that may not be immediately apparent.

Marketing optimization is a growing field – not only in the number of vendors that are offering solutions, but also in the companies embracing ways to enhance data-driven marketing and generate business results. Of the software choices, many of the JavaScript-dependent offerings have inherent risks that are not readily apparent.

Before we get to those risks, let’s briefly explain a little about JavaScript tags. Of the different types of JavaScript tags, there are two main categories: (a) beacons, which collect data and simply report it back to a server, are commonly used by analytics tools and generally do not impact website speed or performance in any meaningful way; (b) tags, commonly used in testing applications that render content function differently than their beacon counterparts and can have negative performance effects that will be shown in the rest of this article.

Here are just a few examples of what your JavaScript-dependent marketing optimization software is hiding from you:

JavaScript tags have site speed implications

There are a number of studies available from companies, as well as independent research firms, that show how site speed affects key business metrics.  For example, Amazon found that just every 0.1 seconds (100 milliseconds) of delay translates to 1 percent of lost sales. Google found that a mere 0.5 seconds (500 milliseconds) of delay resulted in 20 percent loss in traffic. So its no surprise that you would want pages to load quickly, right?

It’s common knowledge that the more JavaScript and general third-party scripts that are on your webpages, the slower that page will be to load for the user. Yottaa, a website speed company, has an insightful blog post that outlines the use of JavaScript on webpages:

“Adding JS has an inherent performance cost and performance risk. Think about the ROI. If it’s not useful for your business goals, don’t keep it on your site just because it’s popular or nice looking.” – Yottaa 

If you conduct a test and page speed is different between your tested version of the page and a new version of the page (even if it’s only 0.1 seconds), user behavior can be affected and test results should be questioned due to the speed implications. Ensure that you are closely monitoring site speed alongside test results to ensure that slower pages are not skewing results.

Tested content could change suddenly

Flicker effect is a common occurrence as a result of testing images. We have all likely seen it and the effect occurs when part of your page is loaded in the visitor’s browser and then overwritten with different content. This could mean that an initial hero image is loaded featuring a child wearing a raincoat and carrying an umbrella, but then another image writes over it showing something different. Images are not the only type of content that has the potential to flicker. Flicker could also be seen in button color, video, or any other on-page element. These types of tag-based artifacts can be disorienting to end users, and introduce noise into test data that can ultimately lead you to make bad decisions. 

Due to the nature of how JavaScript code and tag execution works, the original content (e.g. the hero image) is loaded before the tested content is retrieved from third-party servers, which then replaces the original content. The occurrence of flicker can be influenced by multiple factors such as tag size, network latency, whether the visitor is on a mobile device, and more. 

Flicker effect is noticeable especially if you visit a page that is knowingly being tested, then clear your web browser cookies and refresh that same page in your browser. For example, take a look at these two videos that clearly show flicker effect:

1. In this first example, we see two elements of the page flicker. First the left-hand column text flickers between 5-7 seconds, as it clearly disappears and then comes back. Even more pronounced in this video, though, is the main image which begins to change about 7 seconds into the video and then loads a brand new image.

2. In the second example, we see some text that flickers, and as a result shifts some of the other on-page elements. Watch the text above the blue button at the 13-second mark and you will see this change as well as the button moving. Think about this in terms of you visiting a page, if you were 13-14 seconds into a page visit and something suddenly changed, what would you think?

Tested images load twice, slowing down site speed

When an image is being tested on one of your webpages, regardless of how fast you may think it loads; it will have to be loaded twice. Going back to the flicker effect example, this is because the “hero” image is actually loaded, and then the JavaScript tag code is executed and the new image being tested is loaded to replace it. 

For this post, we looked at five popular online retailers and looked at the file size of their main images: 

If we assume that each of these images is being tested, then the browser is loading up to double the file sizes that are listed. So if it takes 0.5 seconds to load, then with testing (due to double-loading) we can estimate that it will take this process a full second to complete. Given the business implications we shared in the first point – that single extra second from just loading one image can have a huge impact on how users interact with your website and bottom-line business results.

Potential for interrupting portions of your site

As you can see in this YouTube video, due to a Facebook Connect outage on February 8, 2013, was essentially brought down. While your testing software may not bring down your entire site like it did here to NBC News, it may leave your pages that are being tested in an unusable condition with holes in the page where tested content would normally appear. The point is you are vulnerable relying on third-party JavaScript tags on your website to load content: If a service goes down for any reason, then aspects of your site may not function properly.  

Not only could this turn away potential visitors thinking that they may be experiencing technical issues with your site, but it could mean lost revenue and referral traffic as well. 

For example, if you are testing a new “Buy Now” button on your checkout pages, but the testing vendor experiences an outage, then suddenly your entire purchase process grinds to a halt for the people who were segmented into that test until the IT team can go in and physically remove the JavaScript testing code, then test and re-deploy an updated version of the site. In this example of an e-commerce vendor, it means lost revenue and impact on brand recognition given the poor experience.

Requires attention from your IT resources

Your IT team is busy keeping all of the technology, devices and people that interact with them in perfect harmony on a day-in and day-out basis. So working with them in the best and most efficient way is crucial to success for your marketing technology objectives. But by constantly adding tasks and requests to their backlog for adding tags to your site, you are effectively taking time that they could be focused on other enhancements or technology objectives.

Some tag-dependent vendors require a new tag for each element on the page being tested – this has two main effects on your team:

1. It takes away central knowledge of testing and building a culture of optimization from your team, as you are always dependent on your IT group.  

2. It takes away your ability to be agile and implement tests quickly due to lengthy deployment schedules.

In conclusion, be aware of the effects that tags are having on your website and testing performance. Quantify the impacts of these hidden drawbacks and determine if they affect on your bottom line. Even tag-dependent vendors agree that the less content-rendering tags you have on your site, the better you are, so talk to your team and eliminate any unneeded tags from pages.  

About the Author

Jeffrey Vocell is the Product Marketing Manager for SiteSpect, where he’s responsible for strategy and communication of product messaging.   Prior to SiteSpect, Vocell co-founded Trendslide, a mobile business analytics startup. He holds a B.S. in Business Management and Leadership from Daniel Webster College.


Website Magazine Logo
Today's Top Picks for Our Readers:
Recommended by Recommended by NetLine

Leave Your Comment

Login to Comment

Become a Member

Not already a part of our community?
Sign up to participate in the discussion. It's free and quick.

Sign Up


Leave a comment
    Load more comments
    New code