As we know, Google Search Engine takes a lot of factors into consideration to rank websites and calculate search engine results. As a rule, these factors are connected with the webpage itself (URL, quality of the text, meta-data, etc), or the entire website (age of the domain name, number and quality of links, and so on). In 2010, Google took us aback by claiming that the website speed will influence search ranking. That means that now page speed impacts search ranking. But how exactly does speed influence the search ranking? It’s time to clear things out.
In fact, Google didn’t give clear explanations of what “page speed” means, and what’s the connection between speed and ranking. Moreover, company’s representative said that slow mobile websites will even be penalized in search ranking. Apparently, websites providing positive user experience deserve a better promotion in search results than those without proper speed and optimization. But what exactly Google is measuring? Let’s find the answers.
Before we start…
Google was intentionally unclear in stating which certain aspects of page speed have a bigger influence in search ranking. However, they made it clear that content relevancy still retains its utmost importance. Therefore, while can only guess the correlation between speed metrics and search ranking, we can never say that other immeasurable factors are still important. Thus, you should consider the correlations revealed, but keep in mind that this is an only probable influence.
How Was the Research Performed?
Here, we mention the results of research performed by Matt Peters, data scientist at Moz, and guys from Zoompf. To organize the research, specialists made up a list of 2,000 random search queries from 2013 Ranking Factors study. Various samples of queries were taken – some consisting of one word, others being long phrases. Then they figured out top 50 ranked search result URLs for every query and made up a list of 100,000 total pages to check.
To evaluate the speed, the open source tool WebPageTest was used – specialists selected the same browser versions exploited by consumers to get about 40 various performance measurements about how a webpage loads. Chrome was selected for the test, and its cache was cleared to make the results 100% fair.
Over 40 various page metrics for each URL checked were observed, and most of them didn’t show any significant influence on search ranking. It wasn’t surprising. For instance, the number of connections a web browser uses to load a page does not impact search ranking position. In this list, we’ll only mention the metrics that should have influenced the results according to our logic.
Page load time
When measuring page load time, people get into consideration two basic measurements: either fully rendered time, or document complete time. When it comes to documenting complete time, we talk about the time for the page to load before we can interact with it (click buttons, input data). When we mention fully rendered time, we talk about the time it takes to display all images, ads, and analytic trackers (e.g. the background things). You see all these elements when you scroll a page.
Both of these metrics were checked. What was strange is the absence of correlation between these two metrics. It was expected that both of them will influence search ranking: highly ranked pages have a lower document complete or fully rendered time, and vice versa. Yes, page loading was proved to be connected with user experience and conversions, but the clear correlation in this research hasn’t been revealed.
By page size we mean the number of bytes to be downloaded to render a page fully (it includes images, ads, widgets, fonts, and other content). In this case, lack of direct correlation wasn’t a huge surprise. When the specialists checked the median page size for every search rank position, they’ve revealed a correlation that didn’t correspond with their logic.
Considering the results, they’ve made up a theory: lower ranking website belong to minor companies with fewer resources and may have the lower complexity of structure and pages, and less content. While ranking is increasing, so does the complexity. The only exception from this rule are ‘top-dogs’ who have a lot of budget to promote and optimize their websites.
Time to the first byte
When no certain connection between ranking and page load time was revealed, the researchers resolved to check such metrics as the Time to First Byte (TTFB). This metric shows how long it takes your browser to get the first byte of a response from a web server when an URL address is requested. Therefore, this metrics shows the network latency by sending a request to the web server, and the amount of time required for processing and generating a response, as well as the amount to time to send the first byte as the response from the server to your browser.
Again, the results were amazing: the researchers have identified the clear connection between decreasing searching rank results and growing TTFB. Websites with lower TTFB metrics were ranked higher than the ones with a poor TTFB level. When measuring the correlation between speed and ranking, researchers have found out that these metrics had a higher possibility to influence search ranking.
Total image content
After checking the surprising results of the speed-ranking correlation, researchers decided to take a step further and reveal whether the median size of images of web-pages had any influence on search rank position. However, it didn’t bring about any valuable information. As a matter of logic, sites with more images should load slower – there’s more data the transfer. This metric is also closely connected with the fully rendered time we’ve mentioned before. However, no correlation was found.
The data obtained has proved that there is no connection between page load time (regardless of metrics, be that document complete or fully rendered) and Google’s search results. That applies to both generic searches (with 1-2 keywords), and long tail searches (4-5 keywords). Therefore, if the page load time factor does exist in Google algorithm, it’s lost amount a myriad of many other factors.
However, the results have revealed that there’s a correlation between lower time-to-first-byte (TTFB) metrics and better search engine rankings. Websites with a well-developed back-end infrastructure could deliver the content quicker and be ranked better than slower and simpler websites. That means that despite our guesses and logic, back-end performance, not front-end performance has more influence on website’s search engine ranking. But why?
Probably, TTFB can be the easiest metric for Google to estimate. Its robots and crawlers can measure it in a simple way while measuring document complete or fully rendered times requires a full browser. Besides, such metrics as document complete and fully rendered times mostly depend on the capabilities of the browser loading the page as much as on the design, content, and structure of a website. Thus, using TTFB to measure speed and performance can be logical of the Google crawler. Most likely, page rendering time will also be included in the list of factors because it helps to indicate the user experience.
TTFB is not just easy to calculate – this is a great metric to measure the overall performance of the website, because it depends on the following factors:
- The latency between the server and a visitor.
- The level of server loading.
- How quickly the back-end can generate the content.
Websites can decrease the network latency by implementing CDNs (Content Delivery Networks) that boost the speed of website loading regardless of users’ geographical location. So websites can be ranked so highly due to the reason they’re hosted on high capacity websites or are delivered with the help of CDNs, or have optimized database and application layers.
Head for lower TTFB…
Again, two questions stay: are these websites ranked higher because of a better back-end infrastructure, or do they have to improve back-end infrastructure to keep being ranked higher? Both situations are possible, but researchers tend to believe that websites with a well-developed back-end infrastructure get to rank higher, not the opposite.
That can be explained by the fact that the queries with 4-5 search words do not return results for websites with a high traffic. Long-tail searches are typical of smaller sites owned by smaller companies with specific topics that don’t get a large volume of traffic. Therefore, fast websites with lower TTFB are ranked better than the websites with a high TTFB and slow speed.
…but don’t forget about other factors
So, we’ve figured out that back-end performance has a direct impact on search engine ranking. The back-end optimization includes:
- high-capacity web-servers;
- use of CDNs;
- optimization of data queries;
- back-end applications and database servers.
If you want to perform back-end optimization, measure your TTFB with a tool like WebPageTest, and compare it with the results of your competitors.
But while doing it don’t forget that although front-end web performance factors (“document complete” and “fully rendered” times) don’t have a significant influence on search engine ranking, you shouldn’t assume they’re not important for your website. Front-end optimization is about ensuring a great user experience and making your website fast.
Quick website means having more visitors who visit more pages, spend more time there, and come back more often and are more likely to click ads and purchase products. Faster websites mean satisfied visitors, and happy visitors help you to promote your website via links and sharing. All of that improved website searching results. So if you want to perform front-end optimization, get a web performance report – that’s a good starting point for you.
As we can see, TTFB and back-end performance are connected with search engine ranking, while front-end metrics don’t have as much impact. However, the last ones have a direct impact on user experience, so they are also important. If you want to boost search engine rankings as well as make your readers/customers happy, you should care about front-end and back-end performance of your website.