Blog |
Digital Video Playout, google, PageSpeed, seo, mobile
How increasing your Google PageSpeed score could mean a worse video experience for your end users
Thursday, September 27, 2018

Intro to the Value of Page Speed

Google has long established that their SEO algorithm includes page speed as a ranking factor. This was further emphasized in early 2018 when they announced it would be a factor for their mobile index, whereas previously the page speed variable was only applied to the desktop browser index.

Google has provided tools for measuring and identifying problematic areas of websites that impact the page load. While this is certainly an important consumer-centric consideration for any brand, adding SEO impact to the equation incentivizes to brands the need for nimble browser performance.

While the value of speed is undeniable, making pages load fast also means having fewer components that are also valuable for operating in a given website. Social integrations, analytics and ads all negatively impact page speed. Additionally, tools that enable a web publisher’s complex workflow, such as content management systems, inherently add bulk and additional processing required in order to provide their added value to the publisher’s business. After all, nothing is going to be faster than a simple page of text, though that’s generally not what the consumer is looking for, either. As a result, optimizing for speed and responsiveness needs to be balanced with the overall functions needed in any given website. Google’s view (and often rightly so) is that this balance has often fallen far away from good performance in exchange for simplifying (but bulky) frameworks, overly aggressive advertising, social and analytics integrations.

In-Page Video Players and Speed Impacts

As publishers increasingly add video into their content offering, video becomes another factor in the page-speed equation. Video players themselves are relatively light-weight in most cases, but they facilitate a very bulky operation (streaming of segmented or progressive video files). So depending on the desired experience, an embedded video can range from completely lightweight (i.e. not even loaded until a user initiates an action that loads and initializes the player), to integrations that can take up a lot of the page connection’s bandwidth (i.e. preloading ads and content, autoplay of video content, etc). These types of video player configurations may be highly desirable in terms of the desired engagement, but they may work against getting the overall page to load quickly.

Testing Video Player Embed Styles and PageSpeed Insights

At Ooyala, we provide a video player with a wide range of embedding options. This provides significant optionality for our customers so that they can embed as they see fit. However, when considering page speed (specifically, how Google measures it), the way the player is embedded will impact the score differently.

To see how page speed was impacted based on the style of embedding Ooyala’s HTML5 video player, we examined the following embed types:

  1. iFrame
  2. Single Script (Standard embed, not Async*)
  3. Single Script Async (Standard embed, Async*)
  4. Multiple Scripts (Advanced Embed, not Async*)

*Async is short for Asynchronous. In this context, it means that the script that loads the Ooyala player is allowed to load dependencies asynchronously or not.

Additionally, we examined the location of the video player on the page for impacts.

Testing Setup

The player was added to a typical website with video as only a small part of an overall offering on the page (article page with images and video, navigation control, ads and social integrations).

The page was then tested with the different embed types and measured against the time it took to load the player and start video playback (autoplay), as measured from the start of the page load (not the start of the video player). These were then run through Google’s PageSpeed Insights tests to score them. The page itself was not optimized fully and scored a 78 with no video player. This is indicative of a page that certainly has room for improvement, but is also not expected to be in the range of being a significant negative factor in Google's ranking algorithm (based on both statements by Google and on empirical evidence).

Testing Results

The general results were as follows:

Highest Google PageSpeed Insights Score (ranking best to worst):

  1. iFrame
  2. Standard Embed, Async
  3. Advanced Embed, no Async
  4. Standard Embed, no Async

When we looked at the time to video playback (measured from the time the page starts to load), the embed style ranked as follows:

  1. Advanced Embed, no Async
  2. Standard Embed, No Async
  3. Standard Embed, Async
  4. iFrame

As you might notice, the order is almost exactly reversed comparing to the previous test. That is, the higher the embed type scores on the PageSpeed Insights score, the less performance the video player showed in terms of getting the video player running quickly.

We also found that if the video player was embedded below the ‘fold’ (not in the immediate viewable area without scrolling), the video player was excluded from having any impact on the Page Speed Insight scores. While this may be beneficial for video embeds that occur deep within an article page, generally this approach is not going to drive good video engagement.

What it Means

The test results showing that the higher the score of a video player embed type, the worse it performs in terms of video playback speed metrics (i.e. player load, time-to-first-frame) seem counterintuitive at first blush. However, the results are not surprising when you consider the full picture. Google’s Insight Score is not concerned about a video player loading, it’s concerned about the speed of the overall page speed loading, and with the browser having full flexibility to optimize the loading sequence. This is why Google’s Insight Score will identify scripts that do not allow asynchronous loading. However, the video player (whose modular architecture means loading multiple resources before initializing) can get to doing its job faster if the scripts are loaded synchronously. Additionally, iFrame embeds can mean the video is loaded very late relative to all other page elements, giving it staggered, disadvantaged starts.

The bottom line is that while the style of video player embed used can have a moderate impact on your Google Insight score, it can have a major impact on your video startup speed. Which should be the priority in business-specific decisions. Ooyala is dedicated to providing both the options and the insights to enable each publisher to maximize their web performance - whether it be for SEO or Video Engagement metrics.

Steve Paddon

Steve is the Product Lead for OoyalaPLAY, the Ooyala Flex Media Platform application that powers the viewing experience.

You can follow on Twitter and on Linkedin