Search Engine Rankings or Your Users?
Google isn’t completely transparent with how web pages are ranked on their search engine results pages, but Google employee Matt Cutts has in the past confirmed that site speed does have a bearing on search engine ranking. Why? "Faster sites create happy users," Cutts said.
A recent study by Moz (producers of SEO software) carried out by data scientist Matt Peters along with Zoompf (a website performance platform) took an in-depth look at the correlation between site speed and Google search engine rankings.
The findings confirmed that website speed does matter when it comes to search engine rankings, but it’s Time To First Byte (TTFB) that makes the difference rather than page load times.
When building commercial websites, we have to find a happy balance between usability and ensuring the site performs well on search engine rankings: User-friendly SEO.
SEO and usability, side by side, are both important. And, when done correctly and with the best intentions, SEO and usability will not be at odds, and they will actually reinforce one another.
If the conclusion of Peters’ study is true — that TTFB is the measurement used for determining how fast a website is, and thus, how a web page should be ranked in search engine results — this could change the way we go about building fast websites for SEO.
Why the Time To First Byte Study Was Done
To begin the research, Peters worked with the Moz team to create a list of 2,000 random search queries from the Searchmetrics 2013 Ranking Factors Study. They then chose a representative sample of queries with varying amounts of terms (1-5 search terms) and then they took the top 50 ranked results. This enabled the researchers to assemble a list of 100,000 pages to be used in the study.
They then launched 30 Amazon EC2 instances running in the Northern Virginia cloud. These were all loaded with a private version of WebPageTest, a popular website performance and optimization testing tool, to collect more than 40 test measurements.
Google Chrome was used as the test browser and each page was tested with an empty cache.
You can download the raw data from Moz.
The study took two measurements that are commonly associated with site speed. These are:
- Document Complete: the time it takes for a web page to load static content
The researchers expected both of these metrics to have an influence on search engine rankings.
They were taken completely by surprise when no correlation was found.
That then led them to look at Time To First Byte.
Time To First Byte is measured as the time it takes for a web server to send the first byte to the requestor.
TTFB is traditionally a measurement of how responsive a web server is, not how fast and a web page is.
The study revealed "a clear correlation" between low search engine rankings and high TTFB times.
Why Use Time To First Byte?
According to the study, the most likely explanation for Google using TTFB to measure site speed is due to it being the "quickest and easiest metric for Google to capture."
This is because Google’s robots are capable of taking a full measurement of TTFB, whilst gathering document or rendering speeds would require more resources if it were done on a large scale. It’s highly unlikely that Google (or anyone else) would actually have the resources to carry this out, despite the company’s massive data centers, if you consider for a moment the sheer volume of web pages on the Web.
According to the researchers: "Not only is TTFB easy to calculate, but it is also a reasonable metric to gauge the performance of an entire site."
"TTFB is affected by 3 factors," the researchers said. These 3 factors are:
- Network latency
- How heavily loaded the web server is
- The speed in which a site’s backend can push content to the requestor
These factors could be indicative of how fast a web page is, but not in a conclusive way like page loading times can.
Why TTFB is Not Important to Your Users
A study from CloudFlare states that TTFB is irrelevant to the user’s experience. And there is truth to that.
How fast a web page fully loads is all that matters when we begin to look at what is good for our users.
Some things that negatively affect TTFB are actually good for our users, the CloudFlare study discovered. For example, the CloudFlare study saw that when gzip compression was used, the page itself was downloaded 5x faster, but that this negatively affected TTFB by making the time 8x greater.
"From the end user perspective TTFB is almost useless. In this (real) example it’s actually negatively correlated with the download time: the worse the TTFB the better the download time," John Graham-Cumming, a respected programmer in the industry who works for CloudFlare, said.
Therefore, if you care about SEO and your users, in the scenario above, you’re at a crossroads. Do you optimize page speed for your users or TTFB for your search engine rankings?
When it comes to website performance as it relates to the user experience, frontend design is what matters most, and not so much the backend (web servers) which is what TTFB effectively measures.
If a web server can push a byte quickly due to factors such as a shorter distance from the request, server resources, web server architecture, using a CDN, and what web server software you use, it doesn’t automatically mean that the web page being requested is designed optimally to ensure that the rest of the page is quick.
Though there is a good chance that TTFB is also an indicator of web page speed, TTFB is not the best metric to measure it, and there are other better options.
The Bottom Line
It’s important to note that these are just studies and assumptions.
We won’t know for sure whether TTFB is the only measurement that Google uses to determine a web page’s speed or not.
But if it is, then it’s an inadequate way of determining whether a user will have fast and responsive experience on a given web page.
- What Potential Impact Can HTML5 Have on SEO?
- 10 Excellent SEO Tips That Will Improve Your Web Design
- 10 Usability Tips Based on Research Studies
- Related categories: Web Development and Content Strategy
About the Author
This was published on Aug 14, 2013