For years, SEOs have pointed to the importance of page speed and loading times for search engine optimization.
But how important is this ranking factor? When is a website fast enough for Google? What does fast mean? A short loading time, low response time, page size, or download speed?
We have compiled the official information from Google and the results of two own studies on the influence of Page Speed on ranking.
- What does fast mean?
- Page speed benchmark study by Google
- Different types of website performance
- Relationship between “speed” and ranking according to Google
- The Google Speed Update
- Correlation Investigation Page Speed & Ranking (December 2016)
- The questions about the investigation
- Is there a connection between the Google ranking and the point ratings from the Google Page Speed Insights Tool?
- Is there a connection between Google ranking and the loading time?
- Which metrics should you use to optimize page speed from an SEO perspective? Page speed points or loading time?
- Does Google weigh the loading time/page speed more strongly as a ranking factor for a ranking in the top 10 search results?
- How fast do pages load on average and is there a threshold?
- Investigation of the influence of loading time & first-byte time on the ranking (November 2014)
- Further studies on loading time and search engine ranking
- Conclusion: Page Speed is important for UX, little to medium important for SEO
- Tools for testing page speed and loading times
What does fast mean?
There are a lot of assumptions on the Internet when a website is fast. The assumptions range from fast enough for the user to specific times such as no longer than 3 seconds or at least 1.5 seconds. According to Search Console, the time it takes to download a page over 3 seconds seems long. Between 1.5 and 3 seconds on average. According to John Mueller, Google has so far differentiated between slow and fast websites.
With the Page Speed Update in July 2018, Google wants to differentiate more precisely what is fast and slow.
In other words, it does not matter whether someone reaches 85 or 100 points in the Page Speed Insights tool. Google employee Gary Illyes substantiated this statement at least with regard to mobile. So a score of 85 is absolutely sufficient.
In 2016, there was a statement by John Mueller on Twitter, in which he described a speed between two and three seconds as fast. He also points out that the number of http requests is not decisive, but rather the loading time.
From our own investigations with Pingdom, the average loading time of websites is currently around 3.25 seconds (as of December 2016).
It is still important to differentiate the reasons why Google pays attention to loading times and website operators should pay attention to fast websites. Firstly, it is about user experience. Especially on smartphones. For this reason, Google has made page speed a ranking factor. The other is crawling.
According to John Mueller, pages, where the Google bot takes more than 2 seconds to download, are less crawled than fast pages. Google wants to use the crawl budget more effectively. On the other hand, Google emphasizes that the crawl budget is not a problem for many websites. See also Google: Normal-size pages don’t have to worry about the crawl budget.
In this post on the Google Webmaster Central Blog, a Google employee gives the following answers:
As a rule, publishers do not have to worry about the crawl budget described below. If new pages tend to be crawled on the day of their publication, the webmaster doesn’t have to worry about the crawl budget. In addition, a website with fewer than a few thousand URLs is usually crawled efficiently.
Q: Does the speed of the website affect my crawl budget? What about errors?
A: The faster a website is, the better the user experience and the higher the crawl rate. For Googlebot, a fast website is a sign of well-functioning servers. This way, he can access more content over the same number of connections. On the other hand, numerous 5xx errors or timeouts when establishing a connection indicate the opposite and the crawl slows down.
Q: Is crawling a factor in the ranking?
A: A higher crawl frequency does not necessarily lead to better positions in the search results. Google uses hundreds of signals to rank the results. The crawling is necessary to appear in the results, but no ranking signal.
Page speed benchmark study by Google
Google itself published a benchmark study on page speed in 2018. With the following results:
But what time is reasonable for users?
I think this question cannot be answered in general. People who are under time pressure answer this question differently than people who surf the web comfortably.
In addition, it is always a question of the internet connection, how long it takes a website to load. Another factor in feeling speed or slowness is habit. You orientate yourself to what you are used to from other offers.
According to a study by Radware in 2013, the 500 largest e-commerce sites were examined in terms of loading speed. Accordingly, e-commerce websites tend to be slower than faster due to their growing size.
Different types of website performance
Before I go into our investigation results, it makes sense to consider some possible criteria or indicators for the evaluation of the “speed”.
Response time / first-byte time
This metric describes the first contact of a person, so to speak. Browser with the page. Wikipedia:
It is the duration from the virtual user making an HTTP request to the first byte of the page being received by the browser. This time is made up of the socket connection time, the time taken to send the HTTP request, and the time taken to get the first byte of the page.
Page Load Time
The loading time or page load time comprises the complete time of the process between an initial request e.g. Click on a link until the finished rendering of the page. But there are also opinions in which only the process up to the rendering of the page, i.e. loading the Html code without considering the download or the display of images and loading JavaScript code such as Google Analytics, GTM, etc.
Weight or size of the website
The size or weight of a website describes the number of kilobytes it takes to load a website completely (down). In addition to the pure weight, the speed of downloading these kilobytes is also interesting.
Relationship between “speed” and ranking according to Google
Google introduced Page Speed as a new ranking factor in 2010. This video from Maile Ohye from 2010:
Here Maile emphasizes the importance of website speed for users and website operators as well as for Google and that Google has therefore introduced Page Speed as a ranking factor. From the statement that Google Page Speed is important, over the years SEOs made the statement that Page Speed is an important ranking factor.
There is no question that page speed is a ranking factor and there is no question that speed is an important criterion for UX and conversion optimization. This is confirmed by Google and various studies (for example here, here, and here).
But whether the loading time is actually an important ranking factor, how many SEOs claim to be questioned? But before we get to the results of our own investigation, here are some official Google statements on the subject.
In the video above, Maile emphasizes that page speed is a ranking factor among many and that content and relevance are much more important.
“Ranking is a nuanced process and there are over 200 signals, but now speed is one of them. Know that ‘content’ and relevance ‘are still primary. ”, Mail Ohye 2010
When Google speaks of page speed as a ranking factor, it is not entirely clear what it meant. So it says in the official article from 2010:
As part of that effort, today we’re including a new signal in our search ranking algorithms: site speed. Site speed reflects how quickly a website responds to web requests. …We use a variety of sources to determine the speed of a site relative to other sites…
While site speed is a new signal, it doesn’t carry as much weight as the relevance of a page. Currently, fewer than 1% of search queries are affected by the site speed signal in our implementation and the signal for site speed only applies for visitors searching in English on Google.com at this point.
A few months later there was also confirmed in the Webmaster Central Blog:
The website speed indicates how quickly a website responds to web requests … So we decided to take the website speed into account in our search rankings. We use many factors to determine the speed of a website compared to other websites …
Even if the website speed is a new criterion, it is not nearly as important as the relevance of a page. Currently, less than one percent of searches are affected by the site speed criterion in our implementation. The site speed criterion is currently only applied to Google.com searches in English.
2010 and 2011 Matt Cutts still commented on the topic
In the Webmaster Central Hangout from April 8th, 2016, John Mueller confirmed that the influence of server loading time on the ranking was only slight and that optimization in the millisecond range had no effect on the ranking. He only talks about the time it takes a website to load on the server. The time that the Google bot takes to crawl content is not a direct ranking factor, according to John, but only affects the crawling speed for content. Here is the video:
In a video from 2019 from the SEO Mythbusting series, Martin Splitt from Google explains that Google pays attention to performance criteria such as the rendering bits and above, e.g. recognizing the responsiveness of a page. However, nothing is said about the impact of website performance.
The Google Speed Update
Google announced in January 2018 that Page Speed as part of the Mobile-First Index will be a ranking factor in mobile searches. The Google Page Speed Update was then rolled out in July 2018. This made Page Speed a mobile ranking factor. Interestingly, the loading speed also became a quality feature for the quality score for Google AdWords.
However, the impact on the ranking only seems to be as high as for desktop searches, and a distinction is also only made between fast and slow websites.
“The” Speed Update, “as we’re calling it, will only affect pages that deliver the slowest experience to users and will only affect a small percentage of queries. It applies the same standard to all pages, regardless of the technology used to build the page. The intent of the search query is still a very strong signal, so a slow page may still rank highly if it has great, relevant content.
In a webmaster hangout from the beginning of June 2018, however, John Müller explained that with the Mobile Page Speed Update introduced in July, they viewed loading times a little more differently. You should not only make sure that the Google bot gets the page delivered quickly but above all the user. According to its own statement, Google also uses real user data from Google Chrome here.
“Blocking scripts to speed up a page just for Googlebot won’t change how we rate your page.”
John Muller will be even more specific about this in another webmaster hangout. Accordingly, Google uses data from the Chrome User Experience Report for page speed measurement. Users who have given their consent transfer their data to Google. This data is collected and evaluated there. In this way, it can be determined how a website behaves under real conditions.
Also in a hangout on February 5th, 2019, John spoke of Page Speed as a weak ranking factor. The quality of the content is much more important.
This Google video dates from October 2019. Here Martin Splitt and John Mueller give some interesting statements about the question of what is fast and what is slow and which metrics and tools should be used.
The following frequently asked questions are answered:
What is fast and what is slow?
For Google, there is only enough quick and slow classification. There are no clear threshold values to be considered here.
Page Speed Insight, GTMetrics, or Lighthouse. What is the right page speed tool?
Google advises webmasters to pay attention to a mixture of different tools. Each tool has different target groups. GTMetrics is more of a tool for technicians. Other tools are more for marketers. The results from the Page Speed Insight Tool are something in between.
Content Meaningful Paint (CMP) or Time to Interactive (TTI)
As so often, the question of which metrics should be used to optimize the page speed cannot be answered generally. The Content Meaningful Paint (CMP) should be used for pages that have the purpose of informing the user with content. For pages on which a fast interaction e.g. playing a video, performing a download, or filling out a form, on the other hand, the Time to Interactive (TTI) is recommended.
Read Also: Hosted CMS vs. Self-Hosted CMS: Exploring the Advantages and Disadvantages
Correlation Investigation Page Speed & Ranking (December 2016)
Our most recent study on loading times and Google ranking is based on data that we collected in December 2016. We wanted to address the following questions, among others: answer:
- Is there a connection between Google Page Speed points and Google ranking?
- Is there a connection between loading time and ranking
- Are the loading time or page speed points weighted differently for the first 10 search results than for the results from eleven to twenty?
For this investigation, we have compiled the top 20 search results data for 20 keywords with regard to the points according to Google Page Speed (desktop and mobile) and loading time.
The keywords were a colorful mixture of short and mid-tail keywords with transaction-oriented and commercial search intent e.g. the term combinations of content marketing, fashion for chubby, health insurance, disaster protection, olive oil, gift vouchers, time recording, power bank …
We have used the following tools for this:
I then used a correlation analysis to correlate the scores and loading times to the rankings per URI. I then did the same for the loading time in terms of page size and the number of http requests. Since I could not find a tool to query the data in bulk, we had to collect all the data manually, thanks to our hard-working intern Simon! Anyone who has a tool for bulk inquiries at hand always has it. Gladly in the comments …
First of all, the results of the correlation analysis on the relationship between points from the Google Page Speed Tool and the ranking at Google in the Top 20, Top 10, and positions eleven to twenty.
Top 20 Correlation Coefficient (Desktop): -0.40672
Top 10 Correlation Coefficient (Desktop): -0.7694
Correlation Coefficient Top 11-20 (Desktop): -0.06355
Correlation coefficient top 20 (mobile): -0.47139
Correlation coefficient top 10 (mobile): -0.8895
Correlation coefficient top 11-20 (mobile): -0.3946
Correlation coefficient top 20: -0.08871
Correlation coefficient top 10: -0.61247
Correlation coefficient top 11-20: -0.06564
The correlation coefficient
Before I interpret these results, I will briefly explain the correlation coefficients. The correlation coefficient ranges between -1 and +1. It indicates how likely there is a connection between variables or not. With a rather small number of samples, correlation coefficients that range between 0.9 to 1 are considered to be very high. So a connection is very likely.
- up to 0.2 => very low correlation
- up to 0.5 => low correlation
- up to 0.7 => average correlation
- to 0.9 => high correlation
- over 0.9 => very high correlation
With o there is no connection at all. The reverse is true in the minus range. This is important to understand for the investigation based on the points from the Page Speed Insights Tool and the ranking.
Since the assumption is to be examined whether the ranking increases as the number of points increases, the value moves towards position one, the correlation is dysfunctional. The highest possible negative correlation would confirm the presumption. A pronounced correlation coefficient stands for a strong correlation in the loading time.
However, correlation does not mean causality. That is why I interpret the following results in the subjunctive. In addition, every SEO should be aware that the determination of the ranking is so complex today that it is almost impossible to clearly find dependencies between the ranking and individual metrics.
The questions about the investigation
We got a little closer to the answers to the following questions through the investigation:
- Is there a connection between the Google ranking and the point ratings from the Google Page Speed Insights Tool?
- Is there a connection between Google ranking and the loading time?
- Which metrics should you use to optimize page speed from an SEO perspective? Page speed points or loading time?
- Does Google weigh the loading time/page speed more strongly as a ranking factor for a ranking of the top 10 search results?
- Is there a connection between the size of a page and the loading speed?
- Is there a connection between the number of http requests and loading time?
Is there a connection between the Google ranking and the point ratings from the Google Page Speed Insights Tool?
According to the results of our investigation, there appears to be a connection between the rating of the Google Page Speed Tool and the ranking. The correlation between the mobile points and the ranking is greater than that of the desktop points. The correlation coefficient is already high in the top 10 search results at 0.89, which suggests a connection.
However, one must also note that there are always strong outliers. So e.g. my article What is social media marketing? Info, definition, goals, strategy & more.
For years, this has consistently ranked second or third for the term social media marketing. However, with a loading time of over 10-15 seconds, it is already in the very slow range which is uncomfortable for users. And even with the Page Speed Insights Tool, it doesn’t spill glory at all and ranks in the deep red area:
That’s only an example. There are some outliers in almost every data record, which allows a supposition that the weighting of the ranking factor page speed and loading time is maximally mediocre.
That is why we must continue to use Maile Ohye’s statement from 2010 as a guide. There are probably significantly stronger ranking factors than page speed, such as Content, links, and trust/authority of the domain itself.
Is there a connection between Google ranking and the loading time?
The correlation coefficient for the relationship between loading time and Google ranking is rather medium at 0.61 in the top 10 search results. Therefore a connection cannot be ruled out, but it seems to be rather weak.
Which metrics should you use to optimize page speed from an SEO perspective? Page speed points or loading time?
Many page speed optimizers will scream and protest here, but based on the results, I would recommend doing page speed optimization with a view to the mobile points of the page speed insights tool if you want to rank better. But the criticism of this approach would be justified in view of the perceived speed for the user, as Torben Leuschner explains in his article Demystification of the Google PageSpeed Insights Tool by colleagues at ryte:
With one exception (“Average server response time”), Google PageSpeed Insights does not analyze times. So we are talking about the topic “Optimization of loading time and speed” and the time as the most important key figure simply falls under the table.
But with regard to the rankings, these points seem to be a better orientation.
Does Google weigh the loading time/page speed more strongly as a ranking factor for a ranking in the top 10 search results?
Here’s a clear yes. I would even go as far as to say that only for positions one to ten-page speed can be considered as a ranking factor. At least the correlations in positions 11 to 20 are so weak compared to the coefficients in the top 10 that one can assume.
How fast do pages load on average and is there a threshold?
What struck me about the data sets for the different keyword queries is that over 90% of the examined websites loaded faster than 3 seconds. In addition, the average loading time of all websites that were previously measured at Pingdom is currently 3.25 seconds, which would also correspond to what John Mueller said about fast websites.
Investigation of the influence of loading time & first-byte time on the ranking (November 2014)
Our first study on page speed and ranking dated from the end of 2014. In this study, similar to the MOZ study, we checked the influence of the first-byte time, page load time, and weight of a website on the ranking.
For this purpose, we entered 50 search queries from different subject areas on Google and examined the loading time, first-byte time, and weight of the websites ranked in the first ten search results. Here are the results:
Result of the first investigation: only a small influence on the ranking
Our page speed test showed a picture similar to that of the MOZ study (see below). Except for the first-byte time, there does not seem to be a clear direct connection between the speed, loading time, or weight of a website for ranking.
A connection can be guessed at the Firstbyte Time. The esteemed colleague Michael Janssen has made his own investigation, from which the first-byte time of less than 300 ms is given as a recommendation. However, this depends crucially on the speed of the server on which the domain is hosted.
Read Also: Better Understanding the Content Management System
Further studies on loading time and search engine ranking
From further studies in recent years, it is also clear that the “speed” of a website does not appear to be an important ranking signal. In 2013 there was also an investigation by MOZ to what extent the loading time, response time, and the size of a page correlated with the rankings for 2000 search queries.
MOZ was unable to find a correlation between the loading time, the size of the website, and the rankings. The only thing you could guess weakly as a connection to the ranking of a page was the response or first-byte time.
Our data shows there is no correlation between “page load time” (either document complete or fully rendered) and ranking on Google’s search results page. This is true not only for generic searches (one or two keywords) but also for “long tail” searches (4 or 5 keywords) as well.
We did not see websites with faster page load times ranking higher than websites with slower page load times in any consistent fashion. If Page Load Time is a factor in search engine rankings, it is being lost in the noise of other factors.
We had hoped to see some correlation especially for generic one- or two-word queries. Our belief was that the high competition for generic searches would make smaller factors like page speed stand out more. This was not the case.
The Searchmetrics study on the 2015 ranking factors was also unable to establish a connection.
A feature that is strongly based on technology and was present for the first time in the analysis from last year is “Site speed”. This factor was included in the pool of ranking factors for 2013 with the expectation of a clearly positive correlation – however, the correlation for this value, as can be seen in the overview graphic with the changes, was almost “zero”.
Conclusion: Page Speed is important for UX, little to medium important for SEO
Google has been saying it for years and our research suggests it. Page Speed influences the ranking of websites. According to our research results, especially for the top 10 search results. But how much effort should you make if you want to make your website faster? Here you have to ask yourself who you want to do the effort for. For the search engines or the ranking so that the user feels comfortable on the website?
I would recommend primarily to optimize the charging speed for the user. And whoever comes up with the homicide argument User signals on websites influence the ranking must first prove this to me with their own investigation or reference to other investigations. Nobody doubts that Google likes high-performance and user-friendly websites. But there is still a difference between finding good and incorporating it into the ranking algorithm.
John Mueller answered the following to my explicit question about user signals as a direct ranking signal last year:
In my view it makes sense to look at user signals on a larger scale. Especially when we test algorithms, these are things like with any website e.g. A / B testing can do, where you can see what works better, what works less well and it definitely makes sense. If you refer to individual pages, I think it makes less sense that you can include that.
John’s statement is similar to other statements on this topic from Google in the past.
Detailed documentation of all current Google statements and more on this topic in the article Are user signals & user behavior a ranking factor/ranking signal?
That is why the question is still justified why in the SEO world in recent years the (loading) speed of a website has repeatedly been mentioned as an important or even very important ranking signal, although large-scale studies do not and cannot prove this thesis Google itself has not increased the weight of the “speed” ranking signal since the statements made in 2010.
As a critical observer and listener, one more reason to take a closer look at the whole thing with your own small investigations. The feeling has increased since our last survey in 2014. But an important ranking factor? Ranking factor yes! Important or even very important? No!
In my opinion, the topic is always cooked so hot in SEO circles, since it is very technical. And technical SEOs love complicated-looking solutions because they can retain their sovereignty and their exposed position.
Territorial security, so to speak. And SEOs love tactical measures or hacks – implement a measure and immediately better rankings. But in my experience, these times are simply over.
Shaun Anderson from Hobo comes to a similar conclusion, according to his own investigation in his article How Fast Should A Website Load in 2017? summarizes the following:
Is that ranking signal and slight speed improvement justification for spending hours redeveloping your site? On its own – it was not – when you have to choose between SEO clean-up tasks, website redevelopment, content marketing, or making your page more relevant in the first place.
A technical approach to improving user experience, it seems to me, would begin with site speed. A faster site should improve visitor satisfaction levels and the number of conversions, for instance.
This very well might have a second-order effect on your rankings over time – as many SEO think that QUERY COMPLETION & SATISFACTION SIGNALS are significant measurements in how Google orders search engine results pages in 2017.
To be honest, I never expected faster website load times to directly improve the rankings of my website.
There is not much to add!
How to Auto-Optimize Images on WordPress Websites
Tools for testing page speed and loading times
At this point, it should be noted that according to Google there is no tool that measures the measurement of the loading times of a website as Google does internally.
Nevertheless, there are some useful tool resources that can be used to analyze websites for page speed
- Pingdom
- Webpagetest.org
- GMetrix
- Uptrends Tool
- Google Page Speed Insights
- Google Lighthouse
- Bulk Page Speed Test
I hope this article is worthy of you to improve your site’s SEO. Never hesitate to ask any questions related to SEO, Blogging, Digital Marketing, and Affiliate Marketing. We always feel happy to guide the community.