We Analyzed 5 Million Desktop and Mobile Pages. Here’s What We Learned About Page Speed

Introduction

This data-driven study aims to provide a good foundation for the speed of browsing sessions. We look at different metrics and categories to identify patterns or find insights on what works and what doesn’t.

By the end of the study, the reader should be able to gain appreciation on the importance of factors that affect speed performance of a website. We also speak from a viewpoint of the user experience, how these metrics ultimately affect the browsing experience.

The study is structured as follows:

• Part 1 sheds some light on the various page characteristics such as Image Weight that impact speed performance of a website, and rank them according to their importance.

• Part 2 provides a more detailed picture of the various page characteristics by examining some individually. This descriptive information complements and expands on the results presented in Part 1 and helps contextualise patterns and trends.

Methodology

Part 1

The first part of the study aims to answer the following two research questions:

1. How accurately can speed metrics such as Time to First Byte (TTFB) or Start Render be predicted based on page characteristics?

2. What are the most important page characteristics that influence page speed ?

To answer the first question, our approach is based on statistical modelling and machine learning. The method of choice is called Gradient boosted decision trees. Gradient boosting is a popular machine learning technique which is highly scalable, efficient, doesn’t require a lot of data transformations and considered to have a high prediction performance.

The process of training the model follows the commonly used steps: 1) split the data set to train and test sets, 2) fit the model using training data, 3) make predictions for the unseen test set, and finally 4) evaluate the results.

In this study, the prediction accuracy of the model is measured by calculating the correlation between the observed and predicted values of dependent variables (i.e. different speed metrics). Correlation of 0 would indicate that the model with given independent variables (i.e. a page characteristic like Image Weight) is not able to predict a dependent variables (e.g. TTFB) better than randomly guessing, i.e. independent variables don’t contain any relevant information or at least the model is failing to take that information into account. Whereas a maximum value for a perfect fit is 1; independent variables can perfectly explain the observations.

For the second question at hand, the choice was to use a recently developed technique called Leave-One-Feature-Out-Importance (LOFO). The idea behind LOFO is to iteratively remove one independent variable at a time from the data set and measure how much predictive power is lost compared to the full model. If the prediction accuracy is not affected at all, then the feature can be considered to be irrelevant for the task. On the other hand, removing important features should cause large loss of accuracy. The results provide insights into where site owners may need to look for page speed optimizations.

The HttpArchive database acted as datasource. HTTPArchive tracks how the web is built by crawling some 5 Million Webpages pages with Web Page Test, including Lighthouse results, and stores the information in BigQuery where it is publicly available. More information can be found on https://httparchive.org/ and a getting started guide can be accessed here: https://github.com/HTTPArchive/httparchive.org/blob/master/docs/gettingstarted_bigquery.md

We randomly sampled 100.000 rows from the May to July 2019 crawls. Each row contains details about a single page including timings, # of requests, types of requests and sizes. In addition, data points on the # of domains, redirects, errors, https requests, CDN, etc. are available. In total, we looked at 300.000 rows for both Mobile and Desktop. Please note that adding more data points to the model would not change the overall results. We did run the models with 3x more data, but the results were similar to the reported ones.

Part 2

We looked at various speed metrics in more detail that contribute to better user experience. We looked at First Contentful Paint (FCP), First Input Delay (FID), various Image Performance Metrics, and Time to First Byte (TTFB) as our core metrics.

We gathered data from browser sessions for the entire month of May 2019. Here we decided to join the May 2019 HttpArchive data with the June 2019 Chrome-UX data. Chrome-UX data reflects how Chrome users experienced TTFB’s, FCP´s and FID´s in real-world conditions. For example, TTFB data in the HttpArchive has traditionally been measured synthetically in the lab accessed from a single server location, so what makes this dataset unique is that it reflects the real-world server response times experienced by Chrome users as they navigate the web. More information on the Chrome-UX data can be found here: https://developers.google.com/web/tools/chrome-user-experience-report/

Information on defining the SQL queries to extract the desired data were partly drawn from discussions on the HttpArchive. Most of the time we expanded on those to provide novel insights. See: https://discuss.httparchive.org. The SQL queries can be provided on request. One needs to copy-paste those into Google´s BigQuery (see getting started guide above).

We aggregated each metric and sliced it according to various categories. We looked at metrics across devices (Desktop vs Mobile), JavaScript Frameworks, Compression, Use of Third-party Scripts, CMS and Hosting Platforms, CDN Usage and CDN Providers.

We also categorized metrics based on existing benchmarks or provisioned standards. A metric can be categorized mainly as Fast, Average, or Slow. We see that exact numbers make sense, but users will most likely view a browsing experience as a good, average, or poor.

Speed Metrics

To provide some context for the forthcoming sections, please find below a description of the various speed performance metrics.

Overview of Page Speed Metrics

• Time-to-First-Byte (TTFB) is measured as the time from the start of navigation request until the time that the client receives the first byte of the response from the server. It includes network setup time (SSL, DNS, TCP) as well as server-side processing. This metric is useful as it ignores the variability of front end performance and focuses only on network setup and backend response time. Available in both HttpArchive (lab) and Chrome-UX (field) datasets.

• StartRender / First Paint (FP) / First Contentful Paint (FCP) mark the points, immediately after navigation, when the browser renders pixels to the screen. FCP, FP and StartRender are all slightly different but in practice for the vast majority of sites they end up being the same (within measurement error). FCP and first paint are when chrome thinks it painted content. StartRender is observed from the outside and is when the viewport actually changed. StartRender is available in HttpArchive (lab) while FCP in Chrome-UX (field), while the FCP metric is available in both datasets.

• Visually Complete is a user experience metric that identifies the moment in time when users perceive that all the visual elements of a page have completely loaded. Only available in the HttpArchive dataset.

• Speed Index is a performance metric that measures how quickly a page renders visual elements, from the user’s perspective. This is a rate-of-speed metric that is closely related to Visually Complete, which is a moment-in-time measure. Only available in the HttpArchive dataset.

• onLoad calculates the speed when the processing of the page is complete and all the resources on that particular page, such as images, CSS and other functionality have finished downloading. Onload Time is usually integrated with online speed testing tools such as Pingdom. Only available in the HttpArchive dataset.

• Fully Loaded adopts the exact same process to page speed as onLoad but will add an additional two seconds after the Onload trigger has fired to make sure there is no further network activity. The reasoning behind this is to ensure more consistently with tests. Only available in the HttpArchive dataset.

• First Input Delay (FID) measures the time from when a user first interacts with your site (i.e. when they click a link, tap on a button, or use a custom, JavaScript-powered control) to the time when the browser is actually able to respond to that interaction. Only available in the Chrome-UX dataset.

References and learn more http://designingforperformance.com/basics-of-page-speed/ https://addyosmani.com/blog/usability/ https://www.youtube.com/watch?v=XvZ7-Uh0R4Q https://developers.google.com/web/fundamentals/performance/user-centric-performance-metrics https://www.slideshare.net/nicjansma/reliably-measuring-responsiveness

PART 1

As discussed above, in this part of the study we look at the various page characteristics and see how the different speed metrics vary across each. Also, we look how well the selected page characteristics can explain the observed speed metric values.

We consider the following page characteristics:

• Request counts
  • reqTotal
  • reqHtml
  • reqJS
  • reqCSS
  • reqImg
  • reqFont
  • reqFlash
  • reqOther
  • reqAudio
  • reqVideo
  • reqText
  • reqXml
  • reqWebp
  • reqSvg
• Sizes // # of bytes TRANSFERRED (so may be bigger when uncompressed)
  • bytesTotal
  • bytesHtml
  • bytesJS // e.g. average total bytes of JS downloaded per page
  • bytesCSS
  • bytesImg
  • bytesFont
  • bytesFlash
  • bytesOther
  • bytesHtmlDoc // size of the main HTML document
  • bytesAudio
  • bytesVideo
  • bytesText
  • bytesXml
  • bytesWebp
  • bytesSvg
• Script counts
  • num_scripts
  • num_scripts_sync
  • num_scripts_async
  • num_iframes
• Other
  • numDomains // # of unique domains across all requests
  • maxage0 // # of responses with max-age=0
  • gzipTotal // # of bytes xferred for resources that COULD have been gzipped
  • gzipSavings // # of bytes that could have been saved if ALL gzippable resources were gzipped
  • cdn

We consider the following page speed metrics:

• TTFB // time to first byte of HTML doc response
• StartRender // when rendering started
• Visual Complete
• Speed Index // webpagetest.org Speed Index score
• onLoad // window.onload
• Fully Loaded // the page is fully done (according to webpagetest.org)

1. Which Page Characteristics Impact the TTFB Metric?

Figure 1.1

Figure 1.2

Key Takeaways

• Figure 1.1. shows the relatively low correlation between predictions and observed values (about 0.3). This means that the available page characteristics don’t contain all the relevant information to fully explain the observed TTFB values.

• This results are expected as TTFB mostly depends on server-processing factors. Most of the page variables under consideration contain information about on-page rather than server-side processing characteristics (with the exception of the CDN variable).

• The accuracy for Mobile is slightly better than for Desktop.

• Which page characteristics have the largest impact on the TTFB metric? For Desktop, we can conclude that the presence of a CDN has the largest effect on TTFB speed, followed by the weight of the HTMLDoc, the number of domains and the number of HTML requests (see Figure 1.2).

• For Mobile, factors more likely to affect TTFB include the number of HTML requests and - to a lesser extent - the weight of the HTMLDoc weight and CDN availability.

2 Which Page Characteristics Impact the StartRender Metric?

Figure 2.1

Figure 2.2

Key Takeaways

• As we move more to the client side of things, prediction accuracy improves as indicated by the higher correlation numbers for both Desktop and Mobile. More specificially, this means that the selected page charactersitics explain more accurately observed StartRender values (Figure 2.1).

• Accuracy for Mobile appears to be higher than for Desktop.

• For Desktop, pages served by a CDN are indicated to be very important to improve the speed of rendering (Figure 2.2). In addition, the weight of CSS files are also an important page characteristic.

• For Mobile, the picture is somehwat reversed: CSS weight is the most important factor, followed by the availabality of a CDN.

3 Which Page Characteristics Impact the Visual Complete Metric?

Figure 3.1

Figure 3.2

Key Takeaways

• The data on the different page characteristic variables explain the Visual Complete observations reasonably well (Figure 3.1) as the correlations are higher than for TTFB and StartRender (correlations of about 0.7).

• CDN is again an relevant factor for Desktop as Figure 3.2 indicates, although the total weight of a page is as much important for that metric.

• On Mobile, however, total page weight and the number of HTML requests define how well a page scores on the visually complete metric.

4 Which Page Characteristics Impact the Speed Index Metric?

Figure 4.1

Figure 4.2

Key Takeaways

• Accuracy is fairly stable, although correlations dropped to about ~0.6 when compared to the Visual Complete metric (Figure 4.1). Again, Mobile values point to higher predictability than Desktop values.**

• Similiar to Desktop results for the Visual Complete metric, having its own page served via a CDN appears to quickly render visual elements. Somehow less relevant but still important are the number of iframes on a page (Figure 4.2).

• The number of iframes and the weight of CSS files seem to affect mobile rendering, according to the numbers of the SpeedIndex metric.

5 Which Page Characteristics Impact onLoad?

Figure 5.1

Figure 5.2

Key Takeaways

• The available data on page characteristics does inidcates the information to be relevant in explaining observed onLoad values. We see correlations of over 0.8 (Figure 5.1). As before, we detect higher accuracy for Mobile data.

• For Desktop, important website attributes constitute the total weight of a page as well as using a CDN (Figure 5.2).

• At the same time, the total page weight factor is clearly the most important factor for Mobile when it comes to the onLoad metric. The number of HTML requests as another factor follows.

6 Which Page Characteristics Impact Fully Loaded?

Figure 6.1

Figure 6.2

Key Takeaways

• As shown by the very high correlation numbers, the independent page characteristic variables describe almost perfectly the observations for the Fully Loaded metric (Figure 6.1).

• With regards to the importance of single website attributes, we can conclude that the total page weight is the most important feature for Desktop when it comes to processing all the resources on a page. The number of HTML requests ranks second. The same results hold also true for Mobile (Figure 6.2).

PART 2

Section 1: Page Content

1.1 Which JavaScript Framework is the Fastest?

A JavaScript Framework takes the heavy lifting of performing various tasks on the website. Thus, most websites use make use of these frameworks to create pages that are efficient, secure, and standardized. How does a JavaScript Framework affect the page speed of a website?

First Contentful Paint (FCP) measures the time from the page being accessed until the time that the first piece of content is rendered in the page. This gives the user the idea that the page is still loading and is responding.

A slow FCP may give the impression that the page is unresponsive, and may cause the user to refresh the page or close the session, negatively impacting user experience.

We look at the various JavaScript Frameworks and see how the FCP varies across each. We classify FCP into three categories (Fast, Average, Slow) and see the percentage of each category per JS Framework.

• Fast: [0, 1000ms]
• Average: (1000ms, 2500ms]
• Slow: over 2500ms

We also identify the number of websites using each framework, and split the analysis per Page Weight (i.e. Payload Size).

The fastest JavaScript Framework is RightJS, while the slowest is RxJS, with the fastest framework 232% faster than the slowest.

The fastest JavaScript Framework is Wink, while the slowest is Meteor, with the fastest framework 213% faster than the slowest.

The fastest JavaScript Framework is Gatsby, while the slowest is Meteor, with the fastest framework 204% faster than the slowest.

Descriptive Results

We observe the following from this data:

• Small-sized Pages (< 1,264,374 Bytes)
  • 32 out of 32 frameworks are able to maintain a Fast/Average FCP for at least 50% of all sessions
  • 25 of the frameworks have Fast FCP for 48.6% of all sessions
• Medium-sized Pages (between 1,264,374 and 4,019,332 Bytes)
  • 33 out of 33 frameworks are able to maintain a Fast/Average FCP for at least 50% of all sessions
  • 26 of the frameworks have Fast FCP for 43.2% of all sessions
• Large-sized Pages (> 4,019,332 Bytes)
  • 31 out of 31 frameworks are able to maintain a Fast/Average FCP for at least 50% of all sessions
  • 24 of the frameworks have Fast FCP for 40.4% of all sessions

We observe the following from this data:

• Small-sized Pages (< 1,264,374 Bytes)
  • JavaScript Framework files constitute 27.7% to 78.5% of the entire page weight.
  • 9 out of 32 of the JavaScript Framework files take up more than 50% of the entire page weight.
• Medium-sized Pages (between 1,264,374 and 4,019,332 Bytes)
  • JavaScript Framework files constitute 23.3% to 66.5% of the entire page weight.
  • 6 out of 33 of the JavaScript Framework files take up more than 50% of the entire page weight.
• Large-sized Pages (> 4,019,332 Bytes)
  • JavaScript Framework files constitute 9.39% to 36.1% of the entire page weight.
  • 0 out of 31 of the JavaScript Framework files take up more than 50% of the entire page weight.

We observe the following from this data:

• Small-sized Pages (< 1,264,374 Bytes)
  • most widely used framework is React, with 74,127 pages using it, or 25.0% of all pages
• Medium-sized Pages (between 1,264,374 and 4,019,332 Bytes)
  • most widely used framework is React, with 158,454 pages using it, or 53.3% of all pages
• Large-sized Pages (> 4,019,332 Bytes)
  • most widely used framework is React, with 62,377 pages using it, or 21.0% of all pages

The following graph shows us the number of pages that uses each framework across all pages, regardless of Page Weight Category.

Key Takeaways

• JavaScript Frameworks do not scale along with the Page Weight. As the size of a page increases, we see that the percentage of Fast FCP also decreases. The size of the JavaScript Framework does not scale with the size of a page, so for larger-sized pages, we see that the JS Framework comprise a smaller percentage of the entire page weight.

• Adopting the most-widely used frameworks does not guarantee in faster page speed. Looking at the most widely used framework, React, the page speed is actually slower compared to frameworks such as RightJS, Wink, Gatsby. We factored in the page size, and looked at small, medium, and large-sized pages separately to arrive at these conclusions.

1.2 Do Compressing Files Over the Web Improve Page Speed?

Here, we look at how compressing files as they are sent over the web can affect FCP. Compressing files before sending them from the server entails additional work on the browser, as the client needs to decompress the files before rendering them on the browser. However, compressing scripts reduces the page weight, and this can potentially result in faster FCP.

In this study, we look at pages that have implemented compression for brotli, gzip, and deflate formats.

We look at the various levels of compression (in terms of %) and see how the FCP varies across each. We classify FCP into three categories (Fast, Average, Slow) and see the percentage of each category per % of Compressed Scripts.

• Fast: [0, 1000ms]
• Average: (1000ms, 2500ms]
• Slow: over 2500ms

We also identify the number of websites within each Compression Category, and split the analysis per Page Weight (i.e. Payload Size).

Descriptive Results

We observe the following from this data:

• Small-sized Pages (< 880,337 Bytes)
  • 10 out of the 10 compression categories are able to maintan a Fast/Average Page Speed for at least 75% of all sessions
  • 8 compression categories have Fast FCP for 56.3% of the time
• Medium-sized Pages (between 880,337 and 3,625,927 Bytes)
  • 7 out of the 10 compression categories are able to maintan a Fast/Average Page Speed for at least 75% of all sessions
  • 8 compression categories have Fast FCP for 44.6% of the time
• Large-sized Pages (> 3,625,927 Bytes)
  • 6 out of the 10 compression categories are able to maintan a Fast/Average Page Speed for at least 75% of all sessions
  • 8 compression categories have Fast FCP for 40.6% of the time

We observe the following from this data:

• Small-sized Pages (< 880,337 Bytes)
  • pages can have compressed files ranging from 7,978 bytes to 346,689 bytes
• Medium-sized Pages (between 880,337 and 3,625,927 Bytes)
  • pages can have compressed files ranging from 87,786 bytes to 1,613,279 bytes
• Large-sized Pages (> 3,625,927 Bytes)
  • pages can have compressed files ranging from 327,654 bytes to 5,528,444 bytes

We observe the following from this data:

• Small-sized Pages (< 880,337 Bytes)
  • most widely used compression category is [50%, 60%), with 2.2667610^{5} pages using it, or 17.7% of all sessions
• Medium-sized Pages (between 880,337 and 3,625,927 Bytes)
  • most widely used compression category is [40%, 50%), with 4.6676210^{5} pages using it, or 36.5% of all sessions
• Large-sized Pages (> 3,625,927 Bytes)
  • most widely used compression category is [30%, 40%), with 2.2978810^{5} pages using it, or 18.0% of all sessions

The following graph shows us the number of pages that uses each level of compression across all pages, regardless of Page Weight Category.

Key Takeaways

• Smaller-sized pages have a large percentage of sites that perform little to no compression at all. This can be observed by the noticeably high volume of pages in the 0%-10% compression category for smaller pages. However, this behavior is not seen in the middle-sized and large-sized pages.

• Overall, middle-sized and larger-sized pages have a lower compression rate. Interestingly, although smaller pages have a specific segment that performs little to no compression, the overall behavior shows a different behavrior. If we look at the distribution (Fig. 1.2.7 to 1.2.9), more sites tend to compress less parts of the site, as the page weight increases. However, all three page weight categories follow a bell-shaped distribution (except for the 0%-10% for small sites).

• Further studies can be conducted to identify similarities within these small sites, such as the possibility of having micro-sites that are well-developed (i.e. also well-compressed), as there is a rise of E-Commerce Sites, Affiliate Marketing Sites, Blogging and Influencer Sites, and Digital Entrepreneurs.

• Pages that compressed 60%-80% of their pages performed the worst. If we look at the percentage of Fast Session (Fig 1.2.1 to 1.2.3), we will see a dip in this percentage under the 60%-80%. Pages with very low or very high compression have better performances as opposed to a page that has a mix of both.

• Across all pages, we observe that it is more common to compress around 30%-60% of the page.

Given these, compression can be a key factor when looking to improve page speed. A better performance is observed for pages having very low or very high compression. This can be pivotal for large-sized pages aiming to have a greater reach of audience, factoring in download speeds and the speed of the client browser.

1.3 Does the Use of Third-party Scripts Affect Page Speed?

Here, we look at how the use of third-party scripts can affect FCP.

This technical article from Google tells that a Third-party JavaScripts are scripts that are usually added into a site directly from a third-party source. These scripts can include, but are not limited to, advertisements, tracking and analytics, and common functionalities.

Third-party scripts can specifically include:

• Social Media Buttons (e.g. Facebook, Instagram, Twitter, Pinterest)
• Embedded Videos (e.g. YouTube, Vimeo, Flickr, DailyMotion)
• Ads (Frames, Scripts)
• Tracking and Analytics
• A/B Testing Scripts
• Common Use (e.g. Date Formatting, Animation, )

These types of scripts are convenient to use and reduces the time to develop new scripts for existing functionalities. However, this may come at the cost of adding unneccessary bulk to the page and more importantly, limited visibility and governance to the script.

We look at the various levels of usage of third-party scripts (in terms of % of the entire script count) and see how the FCP varies across each. We classify FCP into three categories (Fast, Average, Slow) and see the percentage of each category per % of Third-party Scripts.

We also identify the number of websites within each Script Category, and split the analysis per Page Weight (i.e. Payload Size).

Descriptive Results

We observe the following from this data:

• Small-sized Pages (< 880,346 Bytes)
  • 10 out of 10 categories are able to maintain Fast/Average Page Speed for at least 75% of all sessions
  • 8 categories have Fast FCP for 54.7% of the time
• Medium-sized Pages (between 880,346 and 3,625,754 Bytes)
  • 10 out of 10 categories are able to maintain Fast/Average Page Speed for at least 75% of all sessions
  • 8 categories have Fast FCP for 40.6% of the time
• Large-sized Pages (> 3,625,754 Bytes)
  • 5 out of 10 categories are able to maintain Fast/Average Page Speed for at least 75% of all sessions
  • 8 categories have Fast FCP for 43.8% of the time

We observe the following from this data:

• Small-sized Pages (< 880,346 Bytes)
  • third-party scripts often comprised [0%, 10%) of the page, with 307,308 (23.8%) pages under this category
• Medium-sized Pages (between 880,346 and 3,625,754 Bytes)
  • third-party scripts often comprised [20%, 30%) of the page, with 327,741 (25.4%) pages under this category
• Large-sized Pages (> 3,625,754 Bytes)
  • third-party scripts often comprised [20%, 30%) of the page, with 180,938 (14.0%) pages under this category

The following graph shows us the number of pages along with the % of Third-party Scripts used across all pages, regardless of Page Weight Category.

Key Takeaways

• For small-sized pages, most are using little to none (0%-10%) of third-party scripts.

• For middle-sized and large-sized pages, most are either within the 20%-30% range, or the 90%-100% range. We see two distinct behavior in terms of using third-party sites. While majority of the pages lean towards using 20% to 30% of third-party scripts, there is a sizeable number of sites that are mostly using 90%-100% third-party scripts.

• Pages with a lower percentage of third-party scripts perform better. As the percentage of third-party scripts increase, overall, the page speed becomes worse.

• Sites with 90%-100% of Third-party Scripts do not follow the overall trend in terms of page speed. As pointed out in Section 1.2, there could be a rise in sites that are of a specific design, such as a standard family of sites or a hosting platform that is pre-built and already optimized for a specific purpose. Further studies may be needed to validate this claim.

• It is most common to use third-party scripts that comprises 0%-30% of the entire scripts.

• Each 3rd party script increases page load time by 34.1 milliseconds.

Given these, we note that pages who use a lower percentage of third-party scripts result in better overall performance. We should also note that third-party scripts can impact each website differently (as noted in other studies).

To improve overall page speed, website owners should also check which third-party sites can be a potential cause for slowdown, depending on how they load within their respective sites.

1.4 Image Performance Scores

Images play an important role in website performance, as they take up a sizeable amount of the page weight. Moreover, for pages with images, the user’s attention may be focused on these images. Images should be handled properly to maintain a good user experience.

In this study, we look at four (4) metrics for image performance. These are as follows:

• Use of WebP Images. Developed by Google, WebP is an image format for web that can be smaller in size compared to other file formats, but still results in a similar level of image quality.
• Use of Optimized Images. Images that are optimized have multiple versions of the image to be delivered, based on the circumstances of the user session. Under this category, we include the use of a Content Delivery Network (CDN), Image Compression, and other Image Optimization Web Services.
• Handling for Offscreen Images. Offscreen images include files that are not part of the screen seen by users on the initial load. This means that loading these images may be deferred in order to improve the initial loading and responsiveness of the page.
• Responsive Images Score. The appropriate image should be delivered according to the browser size of the user. Thus, it is bad practice to use an image whose original size is larger than its display size. Image responsiveness also includes dynamic changing of images as the browser is resized (for desktop sessions).

Better handling for images directly ties with improving page speed. We look at how pages are adopting these metrics.

Descriptive Results

We observe that there are pages that have obtained perfect scores for each metric. For each metric, we have the following percentage of websites with perfect scores:

• Uses WebP Images Score: 26.3% out of 3,525,445 websites
• Uses Optimized Images Score: 55.3% out of 4,408,250 websites
• Offscreen Images Score: 64.1% out of 4,463,777 websites
• Responsive Images Score: 72.6% out of 4,648,134 websites

Key Takeaways

• The metric with the highest number of pages with perfect scores is the Responsive Images Score, with 72.6% of all pages having perfect scores. On the other hand, the metric Uses WebP Images Score has the lowest number, with only 26.3% of pages having perfect scores.

• The most adopted metric is the Responsive Images Score, with an average score of 87.1%, while the least adopted metric is the Uses WebP Images Score, with an average score of 59.6%.

• At least 25% of pages have perfect scores for each individual metric, implying that there are pages that aim to provide a better handling for images in pages.

Section 2: User Experience

2.1 First Contentful Paint (FCP) Benchmark Metric

Here, we look at the overall performance for First Contentful Paint (FCP), establishing a benchmark for pages. We separate the metrics for desktop and mobile sessions.

We classify the metric as Fast, Average, or Slow as follows:

• Fast: [0, 1000ms]
• Average: (1000ms, 2500ms]
• Slow: over 2500ms

This is benchmarked from a total of 4,200,766 Desktop and 5,222,592 Mobile pages.

2.2 First Input Delay (FID) Benchmark Metric

First Input Delay (FID) measures the time when a user has made any interaction on a page (such as clicking a link, button, or any custom or default web page controls), to the time that the browser is able to respond to this action.

Here, we look at the overall performance for First Input Delay (FID), establishing a benchmark for pages. We separate the metrics for desktop and mobile sessions.

We classify the metric as Fast, Average, or Slow as follows:

• Fast: [0, 50ms]
• Average: (50ms, 250ms]
• Slow: over 250ms

This is benchmarked from a total of 2,648,856 Desktop and 3,063,111 Mobile pages.

2.3 Benchmark for Other Speed Metrics

Here, we look at other speed metrics aside from FCP and FID.

This is benchmarked from a total of 4,371,973 Desktop and 5,297,442 Mobile pages.

Larger pages take 381% longer to visually load compared to smaller pages.

Larger pages take 486% longer to fully load compared to smaller pages.

Section 3: Content Publishing

3.1 Which Content Management Systems (CMS) Have the Best Performance?

A Content Management System (CMS) is a technology used to manage content online, such as blogs, news, or applications. An example of a CMS is WordPress. We look at all CMS identified by web sessions, and look at how these can possibly affect the page speed.

The Time to First Byte (TTFB) measures the time it takes to receive the first byte on a user’s browser. This factors in the amount of time it takes to (1) send the request, (2) process the request on the server, and (3) receive the response on the user’s browser. Here, we look at TTFB varies across different CMS.

Unlike other metrics, the expected values for TTFB are much lower, since we are focus on the first mark that the request has completed its first back and forth route. We classify TTFB into three categories (Fast, Average, Slow) and see the percentage of each category per CMS.

• Fast: [0, 200ms]
• Average: (200ms, 1000ms]
• Slow: over 1000ms

A lot of factors are being considered when optimizing this metric. This metric focuses on delivery of content throughout the web, wherever a user may be browsing for any page.

We also differentiate among CDN and non-CDN pages, as these can have an impact on delivery in general. We will look at CDN more closely in Section 4.

Descriptive Results

For pages not served by CDN, we observe the following:

• Desktop: 13 out of 20 CMS platforms have Fast TTFB for at least one in three sessions
• Mobile: 9 out of 20 CMS platforms have Fast TTFB for at least one in three sessions

It is interesting to note that WordPress, the most widely used CMS, fails to meet any of these, both for Desktop and Mobile.

Here are aggregate statistics for pages not served by CDN:

• Desktop: Fast (33.6%), Average (43.9%), Slow (22.5%)
• Mobile: Fast (30.6%), Average (45.8%), Slow (23.6%)

The graph above shows the Top 20 Sites, sorted in terms of adoption across pages not served by CDN. To give a better visualization, we excluded WordPress from the plot, as WordPress accounts for 1,676,934 out of 2,334,106 of all pages, or 71.8% of all pages.

WordPress accounts for the following:

• Desktop: 761,560 out of 1,068,671 pages (71.3%)
• Mobile: 915,374 out of 1,265,435 pages (72.3%)

For each of the TTFB categories (Fast, Average, Slow), we gather information on the average values across all CMS platforms. We then looked at the Top 20 CMS Platforms, and counted which of these have above average performance for Fast, Average, and Slow TTFB percentages.

For platforms not in the Top 20, we classified them all under the category Others, and is counted as a single group.

• Desktop: Fast (12 out of 21), Average (7 out of 21), Slow (9 out of 21)
• Mobile: Fast (6 out of 21), Average (7 out of 21), Slow (16 out of 21)

For pages served by CDN, we observe the following:

• Desktop: 7 out of 20 CMS platforms have Fast TTFB for at least one in three sessions
• Mobile: 4 out of 20 CMS platforms have Fast TTFB for at least one in three sessions

Similar to the data above, WordPress also fails to meet satisfactory levels of performance, even for pages served by CDN, both for Desktop and Mobile.

Here are aggregate statistics for pages served by CDN:

• Desktop: Fast (28.3%), Average (45.6%), Slow (26.1%)
• Mobile: Fast (26.6%), Average (42.7%), Slow (30.7%)

The graph above shows the Top 20 Sites, sorted in terms of adoption across pages served by CDN. To give a better visualization, we excluded WordPress from the plot, as WordPress accounts for 261,638 out of 314,136 of all pages, or 83.3% of all pages.

WordPress accounts for the following:

• Desktop: 124,863 out of 150,747 pages (82.8%)
• Mobile: 136,775 out of 163,389 pages (83.7%)

For each of the TTFB categories (Fast, Average, Slow), we gather information on the average values across all CMS platforms. We then looked at the Top 20 CMS Platforms, and counted which of these have above average performance for Fast, Average, and Slow TTFB percentages.

For platforms not in the Top 20, we classified them all under the category Others, and is counted as a single group.

• Desktop: Fast (13 out of 21), Average (7 out of 21), Slow (5 out of 21)
• Mobile: Fast (11 out of 21), Average (10 out of 21), Slow (10 out of 21)

Key Takeaways

• WordPress is widely-used, but has a lackluster performance. WordPress had much lower speeds than the other notable CMS platforms, even when looked under the presence of CDN (Served and Not Served by CDN) and Device Categories (Desktop and Mobile).

• Adobe Experience Manager, Sitefinity, TYPO3 CMS, and Weebly have good performance across all scenarios. They have maintained a Fast page speed in one out of three sessions, whether served by CDN and not served by CDN, as well as for Desktop and Mobile sites.

• On the other hand, Concrete5, Craft CMS, Jimdo, Joomla, Wix, and Wordpress have bad performance across all scenarios. These CMS were not able to maintain a Fast page speed (in one out of three sessions), for any of the cases.

• We also see that performance (for all sessions in the world) is generally better when sites are not served by CDN. This is crucial, as one of the perceived benefits of a CDN is better delivery across the globe. We will explore the impact of each CDN vendor in Section 4.

• Specifically for pages accessed by users in the US Region, the performance based on CMS is either very good, or very poor. Weebly has the best performance, with Fast page loads as high as 60% of the time. Included in this group of high-performing CMS for the US Region are Adobe Experience Manager, DNN, Drupal, Sitefinity, and Squarespace.

• TYPO3 CMS is one of the best performers globally, but does not perform well in the US Region. Looking at this CMS more closely, we see that it has higher adoption among European countries, and the data supports this. TYPO3 CMS has stellar performances within the regions Canada (although not part of Europe), France, Germany, Russia, and UK.

3.2 Time to First Byte (TTFB) by Host

We now look at TTFB measured per Host. Websites take the services of hosting platforms to have their website, along with its contents, become accessible over the web.

Hosts vary in terms of pricing, and even within a host, there are several hosting options, often having varying performance for each. We look at the TTFB metric across various hosts, and see if this can also affect the user experience.

Similar to Section 3.1, we classify TTFB into three categories (Fast, Average, Slow) and see the percentage of each category per CMS.

• Fast: [0, 200ms]
• Average: (200ms, 1000ms]
• Slow: over 1000ms

We also differentiate among CDN and non-CDN pages, as these can have an impact on delivery in general. We will look at CDN more closely in Section 4.

Note: A somewhat similiar graphic has been presented here: https://ismyhostfastyet.com/ . However, these calculations do not take account CDNs, page weight and geographies.

Descriptive Results

For pages not served by CDN, we observe the following:

• Desktop: 15 out of 18 Hosts have Fast TTFB for at least one in three sessions
• Mobile: 13 out of 18 Hosts have Fast TTFB for at least one in three sessions

Here are aggregate statistics for pages not served by CDN:

• Desktop: Fast (37.1%), Average (43.1%), Slow (19.8%)
• Mobile: Fast (33.1%), Average (44.1%), Slow (22.7%)

For each of the TTFB categories (Fast, Average, Slow), we gather information on the average values across all hosting platforms. We then looked at each Host, and counted which of these have above average performance for Fast, Average, and Slow TTFB percentages.

• Desktop: Fast (10 out of 18), Average (5 out of 18), Slow (8 out of 18)
• Mobile: Fast (4 out of 18), Average (6 out of 18), Slow (13 out of 18)

For pages served by CDN, we observe the following:

• Desktop: 9 out of 18 Hosts have Fast TTFB for at least one in three sessions
• Mobile: 4 out of 18 Hosts have Fast TTFB for at least one in three sessions

Here are aggregate statistics for pages served by CDN:

• Desktop: Fast (32.5%), Average (43.7%), Slow (23.8%)
• Mobile: Fast (29.4%), Average (42.9%), Slow (27.7%)

For each of the TTFB categories (Fast, Average, Slow), we gather information on the average values across all hosting platforms. We then looked at each Host, and counted which of these have above average performance for Fast, Average, and Slow TTFB percentages.

• Desktop: Fast (15 out of 18), Average (4 out of 18), Slow (4 out of 18)
• Mobile: Fast (9 out of 18), Average (8 out of 18), Slow (10 out of 18)

Key Takeaways

• Most widely used hosts are Shopify (e-commerce hosting platform), WP Engine (hosting for WordPress), and Siteground (another hosting platform). The first two (Shopify and WP Engine) are performing well. However the third one, Siteground, has noticeably poor performance, even though these three are all managed hosting sites.

• We also notice that GitHub is running mostly on CDN, with very little volume of pages not served by CDN. With their current implementation, they are doing pretty well. They have the highest percentage of Fast page speeds, and also the lowest percentage of Slow page speeds across all scenarios.

• Aside from GitHub, Flywheel and Netlify are able to maintain good performances across all scenarios.

• Wix and Siteground sustained poor performances across all scenarios. Wix, as a hosting platform, does not allow the use of other CMS, hence has its own software implementation that is design-driven. However, this also resulted in poor performance as seen in Section 3.1.

• Shopify has a better overall performance for Mobile, regardless of CDN setup. It is the only hosting platform that maintains a good performance for all Mobile scenarios, but not for Desktop scenarios. As mainly used for e-Commerce, if majority of the sales are coming through mobile purchases, this can be seen as a sensible business move. Currently, sessions (and not purchases) in Shopify are almost equal for Desktop (48.1%) and Mobile (51.9%).

• We also see that performance (for all sessions in the world) is generally better when sites are not served by CDN. We will explore the impact of each CDN vendor in Section 4.

While there are differences in the volume for Mobile and Desktop, we can see that the pages for each host have a distinct characteristic, and is a good factor to consider when looking to optimize TTFB.

Section 4: Content Distribution

4.1 Does a Content Delivery Network (CDN) Improve Page Speed?

Using a Content Delivery Network (CDN) allows page content to be distributed to servers across the globe, for faster access. On top of performance benefits, a CDN also adds a layer of security against some vulnerabilities on a website.

We look at how pages with and without a CDN differ in terms of page speed.

We can see that overall, the pages served by CDN perform worse than pages without CDN. This may seem contrary to the the supposedly faster performance when using a CDN. Given our data, using just any CDN may actually hurt your website’s performance.

In the succeeding section, we look at the different CDN providers, and how the performance varies for each.

4.2 Time to First Byte (TTFB) by CDN provider

In the previous section, we found out that implementing a CDN indicated in an overall worse performance than pages that do not use a CDN. Here, we look at the performance for pages under each CDN Provider. We see that there are differences in TTFB scores within each CDN Provider.

We compare the Fast and Slow performance for each CDN Provider, and compare it with the average. From this, we highlight specific providers that are below global averages.

Key Takeaways

From this, we gather that CDN Providers have varying performances, and that this can impact TTFB. While using a CDN is beneficial, we also have to be aware when choosing our CDN provider.

We have the following summary of performance for each provider:

• Desktop
  • Good (Fast % and Slow % are better than the average across all providers): Airee, Amazon Cloudfront, Azure CDN, CacheFly, EdgeCast, Fastly, GitHub Pages, Google Cloud, KeyCDN, MaxCDN, Netlify
  • Average (Fast % or Slow % is better than the average across all providers):
    CDN77
  • Bad (Fast % and Slow % are worse than the average across all providers):
    Akamai, ArvanCloud, CloudFlare, Fireblade, Incapsula, Sucuri
• Mobile
  • Good (Fast % and Slow % are better than the average across all providers):
    Airee, Amazon Cloudfront, Azure CDN, CDN77, EdgeCast, Fastly, GitHub Pages, Google Cloud, KeyCDN, MaxCDN, Netlify
  • Average (Fast % or Slow % is better than the average across all providers): Fireblade, Incapsula, Sucuri
  • Bad (Fast % and Slow % are worse than the average across all providers):
    Akamai, ArvanCloud, CloudFlare

Section 5: Geographical Distribution

5.1 Which Countries Have the Fastest Page Speed?

How does page speed differ when accessed from a different location? Here, we look at which countries have the fastest page loading times. We identified 11 countries, and observed these in detail:

• Australia
• Brazil
• Canada
• China
• France
• Germany
• India
• Japan
• Russia
• UK
• US

Key Takeaways

We will notice that China has the best performance, most notable for pages accessed without a CDN. Next comes Japan and Germany with fast page speeds above the global averages.

The countries France, UK, Canada, US, and Russia have somewhat average page speed, comparable with global averages.

However, we see Australia, Brazil, and India to have page speeds consistently lower than global averages.

Appendix

In this section, you will find the analysis for Section 3 and Section 4, specific to each area.

6.1 Time to First Byte (TTFB) by CMS, Per Geographical Segment

6.1.1 Time to First Byte (TTFB) by CMS, Australia

General Overview:

• Desktop: 1 out of 20 CMS platforms have Fast/Average TTFB for at least 75% of the time.
• Mobile: 3 out of 20 CMS platforms have Fast/Average TTFB for at least 75% of the time.

Aggregate Statistics:

• Desktop: Fast (19.5%), Average (37.4%), Slow (43.1%)
• Mobile: Fast (20.8%), Average (37.1%), Slow (42.0%)

Page Volume Distribution:

• Desktop: Fast (9 out of 21), Average (8 out of 21), Slow (9 out of 21)
• Mobile: Fast (8 out of 21), Average (3 out of 21), Slow (14 out of 21)

General Overview:

• Desktop: 0 out of 16 CMS platforms have Fast/Average TTFB for at least 75% of the time.
• Mobile: 1 out of 16 CMS platforms have Fast/Average TTFB for at least 75% of the time.

Aggregate Statistics:

• Desktop: Fast (18.3%), Average (35.1%), Slow (46.6%)
• Mobile: Fast (19.8%), Average (32.7%), Slow (47.5%)

Page Volume Distribution:

• Desktop: Fast (11 out of 21), Average (9 out of 21), Slow (8 out of 21)
• Mobile: Fast (9 out of 21), Average (6 out of 21), Slow (7 out of 21)

6.1.2 Time to First Byte (TTFB) by CMS, Brazil

General Overview:

• Desktop: 0 out of 20 CMS platforms have Fast/Average TTFB for at least 75% of the time.
• Mobile: 0 out of 20 CMS platforms have Fast/Average TTFB for at least 75% of the time.

Aggregate Statistics:

• Desktop: Fast (20.2%), Average (41.2%), Slow (38.6%)
• Mobile: Fast (18.6%), Average (37.6%), Slow (43.9%)

Page Volume Distribution:

• Desktop: Fast (10 out of 21), Average (8 out of 21), Slow (7 out of 21)
• Mobile: Fast (8 out of 21), Average (5 out of 21), Slow (12 out of 21)

General Overview:

• Desktop: 0 out of 18 CMS platforms have Fast/Average TTFB for at least 75% of the time.
• Mobile: 1 out of 18 CMS platforms have Fast/Average TTFB for at least 75% of the time.

Aggregate Statistics:

• Desktop: Fast (16.6%), Average (39.2%), Slow (44.2%)
• Mobile: Fast (17.1%), Average (32.9%), Slow (50.0%)

Page Volume Distribution:

• Desktop: Fast (14 out of 21), Average (10 out of 21), Slow (6 out of 21)
• Mobile: Fast (10 out of 21), Average (5 out of 21), Slow (11 out of 21)

6.1.3 Time to First Byte (TTFB) by CMS, Canada

General Overview:

• Desktop: 9 out of 20 CMS platforms have Fast/Average TTFB for at least 75% of the time.
• Mobile: 9 out of 20 CMS platforms have Fast/Average TTFB for at least 75% of the time.

Aggregate Statistics:

• Desktop: Fast (32.3%), Average (45.9%), Slow (21.8%)
• Mobile: Fast (30.5%), Average (47.1%), Slow (22.4%)

Page Volume Distribution:

• Desktop: Fast (10 out of 21), Average (9 out of 21), Slow (10 out of 21)
• Mobile: Fast (5 out of 21), Average (5 out of 21), Slow (14 out of 21)

General Overview:

• Desktop: 4 out of 17 CMS platforms have Fast/Average TTFB for at least 75% of the time.
• Mobile: 5 out of 17 CMS platforms have Fast/Average TTFB for at least 75% of the time.

Aggregate Statistics:

• Desktop: Fast (27.4%), Average (43.8%), Slow (28.7%)
• Mobile: Fast (25.8%), Average (43.7%), Slow (30.4%)

Page Volume Distribution:

• Desktop: Fast (12 out of 21), Average (9 out of 21), Slow (4 out of 21)
• Mobile: Fast (10 out of 21), Average (10 out of 21), Slow (6 out of 21)

6.1.4 Time to First Byte (TTFB) by CMS, China

General Overview:

• Desktop: 7 out of 15 CMS platforms have Fast/Average TTFB for at least 75% of the time.
• Mobile: 3 out of 8 CMS platforms have Fast/Average TTFB for at least 75% of the time.

Aggregate Statistics:

• Desktop: Fast (28.4%), Average (35.7%), Slow (35.9%)
• Mobile: Fast (30.5%), Average (43.6%), Slow (25.9%)

Page Volume Distribution:

• Desktop: Fast (9 out of 21), Average (8 out of 21), Slow (7 out of 21)
• Mobile: Fast (3 out of 21), Average (1 out of 21), Slow (2 out of 21)

General Overview:

• Desktop: 1 out of 9 CMS platforms have Fast/Average TTFB for at least 75% of the time.
• Mobile: 2 out of 5 CMS platforms have Fast/Average TTFB for at least 75% of the time.

Aggregate Statistics:

• Desktop: Fast (21.4%), Average (36.4%), Slow (42.2%)
• Mobile: Fast (37.1%), Average (34.6%), Slow (28.4%)

Page Volume Distribution:

• Desktop: Fast (10 out of 21), Average (7 out of 21), Slow (6 out of 21)
• Mobile: Fast (2 out of 21), Average (2 out of 21), Slow (2 out of 21)

6.1.5 Time to First Byte (TTFB) by CMS, France

General Overview:

• Desktop: 10 out of 20 CMS platforms have Fast/Average TTFB for at least 75% of the time.
• Mobile: 8 out of 20 CMS platforms have Fast/Average TTFB for at least 75% of the time.

Aggregate Statistics:

• Desktop: Fast (31.3%), Average (46.3%), Slow (22.4%)
• Mobile: Fast (30.7%), Average (47.9%), Slow (21.5%)

Page Volume Distribution:

• Desktop: Fast (12 out of 21), Average (8 out of 21), Slow (12 out of 21)
• Mobile: Fast (10 out of 21), Average (5 out of 21), Slow (11 out of 21)

General Overview:

• Desktop: 9 out of 18 CMS platforms have Fast/Average TTFB for at least 75% of the time.
• Mobile: 6 out of 19 CMS platforms have Fast/Average TTFB for at least 75% of the time.

Aggregate Statistics:

• Desktop: Fast (31.5%), Average (43.3%), Slow (25.2%)
• Mobile: Fast (31.3%), Average (43.8%), Slow (24.9%)

Page Volume Distribution:

• Desktop: Fast (12 out of 21), Average (11 out of 21), Slow (7 out of 21)
• Mobile: Fast (10 out of 21), Average (10 out of 21), Slow (9 out of 21)

6.1.6 Time to First Byte (TTFB) by CMS, Germany

General Overview:

• Desktop: 7 out of 20 CMS platforms have Fast/Average TTFB for at least 75% of the time.
• Mobile: 9 out of 20 CMS platforms have Fast/Average TTFB for at least 75% of the time.

Aggregate Statistics:

• Desktop: Fast (32.0%), Average (46.9%), Slow (21.1%)
• Mobile: Fast (32.7%), Average (48.1%), Slow (19.1%)

Page Volume Distribution:

• Desktop: Fast (8 out of 21), Average (7 out of 21), Slow (11 out of 21)
• Mobile: Fast (12 out of 21), Average (6 out of 21), Slow (13 out of 21)

General Overview:

• Desktop: 9 out of 20 CMS platforms have Fast/Average TTFB for at least 75% of the time.
• Mobile: 12 out of 20 CMS platforms have Fast/Average TTFB for at least 75% of the time.

Aggregate Statistics:

• Desktop: Fast (31.9%), Average (44.6%), Slow (23.5%)
• Mobile: Fast (32.5%), Average (43.3%), Slow (24.3%)

Page Volume Distribution:

• Desktop: Fast (9 out of 21), Average (10 out of 21), Slow (6 out of 21)
• Mobile: Fast (12 out of 21), Average (8 out of 21), Slow (10 out of 21)

6.1.7 Time to First Byte (TTFB) by CMS, India

General Overview:

• Desktop: 0 out of 20 CMS platforms have Fast/Average TTFB for at least 75% of the time.
• Mobile: 0 out of 20 CMS platforms have Fast/Average TTFB for at least 75% of the time.

Aggregate Statistics:

• Desktop: Fast (17.1%), Average (36.3%), Slow (46.6%)
• Mobile: Fast (15.1%), Average (29.4%), Slow (55.4%)

Page Volume Distribution:

• Desktop: Fast (10 out of 21), Average (8 out of 21), Slow (8 out of 21)
• Mobile: Fast (8 out of 21), Average (2 out of 21), Slow (13 out of 21)

General Overview:

• Desktop: 0 out of 16 CMS platforms have Fast/Average TTFB for at least 75% of the time.
• Mobile: 0 out of 17 CMS platforms have Fast/Average TTFB for at least 75% of the time.

Aggregate Statistics:

• Desktop: Fast (14.7%), Average (32.7%), Slow (52.6%)
• Mobile: Fast (14.6%), Average (25.7%), Slow (59.7%)

Page Volume Distribution:

• Desktop: Fast (14 out of 21), Average (15 out of 21), Slow (4 out of 21)
• Mobile: Fast (8 out of 21), Average (7 out of 21), Slow (9 out of 21)

6.1.8 Time to First Byte (TTFB) by CMS, Japan

General Overview:

• Desktop: 3 out of 20 CMS platforms have Fast/Average TTFB for at least 75% of the time.
• Mobile: 2 out of 20 CMS platforms have Fast/Average TTFB for at least 75% of the time.

Aggregate Statistics:

• Desktop: Fast (23.7%), Average (42.7%), Slow (33.6%)
• Mobile: Fast (24.3%), Average (43.1%), Slow (32.6%)

Page Volume Distribution:

• Desktop: Fast (11 out of 21), Average (7 out of 21), Slow (8 out of 21)
• Mobile: Fast (8 out of 21), Average (2 out of 21), Slow (12 out of 21)

General Overview:

• Desktop: 2 out of 16 CMS platforms have Fast/Average TTFB for at least 75% of the time.
• Mobile: 2 out of 15 CMS platforms have Fast/Average TTFB for at least 75% of the time.

Aggregate Statistics:

• Desktop: Fast (22.8%), Average (40.1%), Slow (37.0%)
• Mobile: Fast (24.3%), Average (35.9%), Slow (39.8%)

Page Volume Distribution:

• Desktop: Fast (11 out of 21), Average (10 out of 21), Slow (7 out of 21)
• Mobile: Fast (8 out of 21), Average (8 out of 21), Slow (6 out of 21)

6.1.9 Time to First Byte (TTFB) by CMS, Russia

General Overview:

• Desktop: 7 out of 19 CMS platforms have Fast/Average TTFB for at least 75% of the time.
• Mobile: 3 out of 20 CMS platforms have Fast/Average TTFB for at least 75% of the time.

Aggregate Statistics:

• Desktop: Fast (28.4%), Average (47.3%), Slow (24.4%)
• Mobile: Fast (24.3%), Average (51.0%), Slow (24.8%)

Page Volume Distribution:

• Desktop: Fast (10 out of 21), Average (11 out of 21), Slow (9 out of 21)
• Mobile: Fast (7 out of 21), Average (6 out of 21), Slow (14 out of 21)

General Overview:

• Desktop: 3 out of 17 CMS platforms have Fast/Average TTFB for at least 75% of the time.
• Mobile: 3 out of 17 CMS platforms have Fast/Average TTFB for at least 75% of the time.

Aggregate Statistics:

• Desktop: Fast (24.8%), Average (47.7%), Slow (27.5%)
• Mobile: Fast (23.2%), Average (45.4%), Slow (31.4%)

Page Volume Distribution:

• Desktop: Fast (16 out of 21), Average (10 out of 21), Slow (4 out of 21)
• Mobile: Fast (10 out of 21), Average (10 out of 21), Slow (5 out of 21)

6.1.10 Time to First Byte (TTFB) by CMS, UK

General Overview:

• Desktop: 12 out of 20 CMS platforms have Fast/Average TTFB for at least 75% of the time.
• Mobile: 10 out of 20 CMS platforms have Fast/Average TTFB for at least 75% of the time.

Aggregate Statistics:

• Desktop: Fast (33.4%), Average (44.8%), Slow (21.8%)
• Mobile: Fast (31.6%), Average (46.0%), Slow (22.5%)

Page Volume Distribution:

• Desktop: Fast (12 out of 21), Average (9 out of 21), Slow (11 out of 21)
• Mobile: Fast (7 out of 21), Average (5 out of 21), Slow (12 out of 21)

General Overview:

• Desktop: 6 out of 19 CMS platforms have Fast/Average TTFB for at least 75% of the time.
• Mobile: 6 out of 18 CMS platforms have Fast/Average TTFB for at least 75% of the time.

Aggregate Statistics:

• Desktop: Fast (27.6%), Average (41.7%), Slow (30.8%)
• Mobile: Fast (28.4%), Average (43.1%), Slow (28.5%)

Page Volume Distribution:

• Desktop: Fast (15 out of 21), Average (11 out of 21), Slow (2 out of 21)
• Mobile: Fast (11 out of 21), Average (8 out of 21), Slow (9 out of 21)

6.1.11 Time to First Byte (TTFB) by CMS, US

General Overview:

• Desktop: 7 out of 20 CMS platforms have Fast/Average TTFB for at least 75% of the time.
• Mobile: 6 out of 20 CMS platforms have Fast/Average TTFB for at least 75% of the time.

Aggregate Statistics:

• Desktop: Fast (31.3%), Average (45.3%), Slow (23.4%)
• Mobile: Fast (28.6%), Average (45.5%), Slow (25.9%)

Page Volume Distribution:

• Desktop: Fast (8 out of 21), Average (8 out of 21), Slow (10 out of 21)
• Mobile: Fast (10 out of 21), Average (9 out of 21), Slow (11 out of 21)

General Overview:

• Desktop: 5 out of 20 CMS platforms have Fast/Average TTFB for at least 75% of the time.
• Mobile: 5 out of 20 CMS platforms have Fast/Average TTFB for at least 75% of the time.

Aggregate Statistics:

• Desktop: Fast (28.1%), Average (48.4%), Slow (23.5%)
• Mobile: Fast (26.2%), Average (46.4%), Slow (27.3%)

Page Volume Distribution:

• Desktop: Fast (11 out of 21), Average (6 out of 21), Slow (10 out of 21)
• Mobile: Fast (11 out of 21), Average (8 out of 21), Slow (9 out of 21)

6.2 Time to First Byte (TTFB) by Host, Per Geographical Segment

6.2.1 Time to First Byte (TTFB) by Host, Australia

General Overview:

• Desktop: 3 out of 18 Host have Fast/Average TTFB for at least 75% of the time.
• Mobile: 2 out of 18 Host have Fast/Average TTFB for at least 75% of the time.

Aggregate Statistics:

• Desktop: Fast (22.1%), Average (40.5%), Slow (37.3%)
• Mobile: Fast (22.7%), Average (36.8%), Slow (40.5%)

Page Volume Distribution:

• Desktop: Fast (7 out of 18), Average (7 out of 18), Slow (10 out of 18)
• Mobile: Fast (5 out of 18), Average (9 out of 18), Slow (10 out of 18)

General Overview:

• Desktop: 2 out of 16 Host have Fast/Average TTFB for at least 75% of the time.
• Mobile: 2 out of 16 Host have Fast/Average TTFB for at least 75% of the time.

Aggregate Statistics:

• Desktop: Fast (21.5%), Average (37.0%), Slow (41.5%)
• Mobile: Fast (21.4%), Average (35.6%), Slow (43.1%)

Page Volume Distribution:

• Desktop: Fast (8 out of 16), Average (12 out of 16), Slow (7 out of 16)
• Mobile: Fast (6 out of 16), Average (9 out of 16), Slow (9 out of 16)

6.2.2 Time to First Byte (TTFB) by Host, Brazil

General Overview:

• Desktop: 2 out of 18 Host have Fast/Average TTFB for at least 75% of the time.
• Mobile: 1 out of 17 Host have Fast/Average TTFB for at least 75% of the time.

Aggregate Statistics:

• Desktop: Fast (22.7%), Average (45.0%), Slow (32.3%)
• Mobile: Fast (21.9%), Average (38.7%), Slow (39.4%)

Page Volume Distribution:

• Desktop: Fast (10 out of 18), Average (8 out of 18), Slow (11 out of 18)
• Mobile: Fast (8 out of 17), Average (9 out of 17), Slow (8 out of 17)

General Overview:

• Desktop: 2 out of 16 Host have Fast/Average TTFB for at least 75% of the time.
• Mobile: 2 out of 16 Host have Fast/Average TTFB for at least 75% of the time.

Aggregate Statistics:

• Desktop: Fast (23.7%), Average (43.3%), Slow (33.0%)
• Mobile: Fast (22.1%), Average (36.9%), Slow (41.0%)

Page Volume Distribution:

• Desktop: Fast (8 out of 16), Average (9 out of 16), Slow (9 out of 16)
• Mobile: Fast (8 out of 16), Average (9 out of 16), Slow (8 out of 16)

6.2.3 Time to First Byte (TTFB) by Host, Canada

General Overview:

• Desktop: 13 out of 18 Host have Fast/Average TTFB for at least 75% of the time.
• Mobile: 13 out of 18 Host have Fast/Average TTFB for at least 75% of the time.

Aggregate Statistics:

• Desktop: Fast (39.8%), Average (44.3%), Slow (15.9%)
• Mobile: Fast (37.5%), Average (45.2%), Slow (17.2%)

Page Volume Distribution:

• Desktop: Fast (7 out of 18), Average (9 out of 18), Slow (11 out of 18)
• Mobile: Fast (7 out of 18), Average (9 out of 18), Slow (9 out of 18)

General Overview:

• Desktop: 10 out of 16 Host have Fast/Average TTFB for at least 75% of the time.
• Mobile: 8 out of 16 Host have Fast/Average TTFB for at least 75% of the time.

Aggregate Statistics:

• Desktop: Fast (37.0%), Average (41.3%), Slow (21.8%)
• Mobile: Fast (34.2%), Average (43.8%), Slow (22.0%)

Page Volume Distribution:

• Desktop: Fast (11 out of 16), Average (11 out of 16), Slow (2 out of 16)
• Mobile: Fast (7 out of 16), Average (9 out of 16), Slow (5 out of 16)

6.2.4 Time to First Byte (TTFB) by Host, China

General Overview:

• Desktop: 1 out of 14 Host have Fast/Average TTFB for at least 75% of the time.
• Mobile: 0 out of 2 Host have Fast/Average TTFB for at least 75% of the time.

Aggregate Statistics:

• Desktop: Fast (14.9%), Average (37.0%), Slow (48.1%)
• Mobile: Fast (27.5%), Average (47.1%), Slow (25.4%)

Page Volume Distribution:

• Desktop: Fast (7 out of 14), Average (8 out of 14), Slow (5 out of 14)
• Mobile: Fast (2 out of 2), Average (0 out of 2), Slow (0 out of 2)

General Overview:

• Desktop: 2 out of 12 Host have Fast/Average TTFB for at least 75% of the time.
• Mobile: 2 out of 2 Host have Fast/Average TTFB for at least 75% of the time.

Aggregate Statistics:

• Desktop: Fast (21.9%), Average (40.9%), Slow (37.2%)
• Mobile: Fast (45.1%), Average (36.1%), Slow (18.8%)

Page Volume Distribution:

• Desktop: Fast (2 out of 12), Average (6 out of 12), Slow (10 out of 12)
• Mobile: Fast (1 out of 2), Average (1 out of 2), Slow (1 out of 2)

6.2.5 Time to First Byte (TTFB) by Host, France

General Overview:

• Desktop: 10 out of 18 Host have Fast/Average TTFB for at least 75% of the time.
• Mobile: 8 out of 18 Host have Fast/Average TTFB for at least 75% of the time.

Aggregate Statistics:

• Desktop: Fast (33.9%), Average (45.9%), Slow (20.1%)
• Mobile: Fast (32.7%), Average (46.3%), Slow (21.0%)

Page Volume Distribution:

• Desktop: Fast (7 out of 18), Average (7 out of 18), Slow (10 out of 18)
• Mobile: Fast (5 out of 18), Average (9 out of 18), Slow (9 out of 18)

General Overview:

• Desktop: 8 out of 16 Host have Fast/Average TTFB for at least 75% of the time.
• Mobile: 5 out of 16 Host have Fast/Average TTFB for at least 75% of the time.

Aggregate Statistics:

• Desktop: Fast (34.4%), Average (43.6%), Slow (22.1%)
• Mobile: Fast (31.8%), Average (47.2%), Slow (21.1%)

Page Volume Distribution:

• Desktop: Fast (7 out of 16), Average (11 out of 16), Slow (7 out of 16)
• Mobile: Fast (6 out of 16), Average (7 out of 16), Slow (9 out of 16)

6.2.6 Time to First Byte (TTFB) by Host, Germany

General Overview:

• Desktop: 8 out of 18 Host have Fast/Average TTFB for at least 75% of the time.
• Mobile: 9 out of 18 Host have Fast/Average TTFB for at least 75% of the time.

Aggregate Statistics:

• Desktop: Fast (36.0%), Average (45.5%), Slow (18.5%)
• Mobile: Fast (34.5%), Average (45.9%), Slow (19.6%)

Page Volume Distribution:

• Desktop: Fast (5 out of 18), Average (8 out of 18), Slow (11 out of 18)
• Mobile: Fast (5 out of 18), Average (6 out of 18), Slow (9 out of 18)

General Overview:

• Desktop: 6 out of 17 Host have Fast/Average TTFB for at least 75% of the time.
• Mobile: 5 out of 16 Host have Fast/Average TTFB for at least 75% of the time.

Aggregate Statistics:

• Desktop: Fast (33.0%), Average (43.1%), Slow (23.9%)
• Mobile: Fast (32.8%), Average (43.9%), Slow (23.3%)

Page Volume Distribution:

• Desktop: Fast (8 out of 17), Average (12 out of 17), Slow (4 out of 17)
• Mobile: Fast (6 out of 16), Average (8 out of 16), Slow (6 out of 16)

6.2.7 Time to First Byte (TTFB) by Host, India

General Overview:

• Desktop: 1 out of 18 Host have Fast/Average TTFB for at least 75% of the time.
• Mobile: 1 out of 18 Host have Fast/Average TTFB for at least 75% of the time.

Aggregate Statistics:

• Desktop: Fast (19.7%), Average (38.3%), Slow (42.0%)
• Mobile: Fast (16.1%), Average (30.6%), Slow (53.3%)

Page Volume Distribution:

• Desktop: Fast (7 out of 18), Average (9 out of 18), Slow (9 out of 18)
• Mobile: Fast (5 out of 18), Average (6 out of 18), Slow (10 out of 18)

General Overview:

• Desktop: 2 out of 15 Host have Fast/Average TTFB for at least 75% of the time.
• Mobile: 0 out of 15 Host have Fast/Average TTFB for at least 75% of the time.

Aggregate Statistics:

• Desktop: Fast (19.1%), Average (33.1%), Slow (47.8%)
• Mobile: Fast (16.3%), Average (28.5%), Slow (55.3%)

Page Volume Distribution:

• Desktop: Fast (7 out of 15), Average (12 out of 15), Slow (7 out of 15)
• Mobile: Fast (5 out of 15), Average (6 out of 15), Slow (8 out of 15)

6.2.8 Time to First Byte (TTFB) by Host, Japan

General Overview:

• Desktop: 3 out of 18 Host have Fast/Average TTFB for at least 75% of the time.
• Mobile: 4 out of 18 Host have Fast/Average TTFB for at least 75% of the time.

Aggregate Statistics:

• Desktop: Fast (27.5%), Average (45.0%), Slow (27.6%)
• Mobile: Fast (26.1%), Average (40.8%), Slow (33.1%)

Page Volume Distribution:

• Desktop: Fast (6 out of 18), Average (9 out of 18), Slow (9 out of 18)
• Mobile: Fast (6 out of 18), Average (10 out of 18), Slow (7 out of 18)

General Overview:

• Desktop: 3 out of 15 Host have Fast/Average TTFB for at least 75% of the time.
• Mobile: 4 out of 14 Host have Fast/Average TTFB for at least 75% of the time.

Aggregate Statistics:

• Desktop: Fast (25.3%), Average (42.8%), Slow (31.9%)
• Mobile: Fast (25.2%), Average (41.5%), Slow (33.4%)

Page Volume Distribution:

• Desktop: Fast (9 out of 15), Average (10 out of 15), Slow (6 out of 15)
• Mobile: Fast (7 out of 14), Average (9 out of 14), Slow (7 out of 14)

6.2.9 Time to First Byte (TTFB) by Host, Russia

General Overview:

• Desktop: 3 out of 18 Host have Fast/Average TTFB for at least 75% of the time.
• Mobile: 4 out of 18 Host have Fast/Average TTFB for at least 75% of the time.

Aggregate Statistics:

• Desktop: Fast (27.7%), Average (48.3%), Slow (24.0%)
• Mobile: Fast (26.5%), Average (47.7%), Slow (25.8%)

Page Volume Distribution:

• Desktop: Fast (9 out of 18), Average (9 out of 18), Slow (10 out of 18)
• Mobile: Fast (7 out of 18), Average (6 out of 18), Slow (9 out of 18)

General Overview:

• Desktop: 3 out of 15 Host have Fast/Average TTFB for at least 75% of the time.
• Mobile: 4 out of 15 Host have Fast/Average TTFB for at least 75% of the time.

Aggregate Statistics:

• Desktop: Fast (26.5%), Average (46.2%), Slow (27.3%)
• Mobile: Fast (25.2%), Average (43.8%), Slow (31.0%)

Page Volume Distribution:

• Desktop: Fast (10 out of 15), Average (10 out of 15), Slow (8 out of 15)
• Mobile: Fast (7 out of 15), Average (9 out of 15), Slow (7 out of 15)

6.2.10 Time to First Byte (TTFB) by Host, UK

General Overview:

• Desktop: 11 out of 18 Host have Fast/Average TTFB for at least 75% of the time.
• Mobile: 10 out of 18 Host have Fast/Average TTFB for at least 75% of the time.

Aggregate Statistics:

• Desktop: Fast (37.2%), Average (45.4%), Slow (17.4%)
• Mobile: Fast (34.9%), Average (46.6%), Slow (18.5%)

Page Volume Distribution:

• Desktop: Fast (7 out of 18), Average (8 out of 18), Slow (10 out of 18)
• Mobile: Fast (7 out of 18), Average (9 out of 18), Slow (10 out of 18)

General Overview:

• Desktop: 9 out of 17 Host have Fast/Average TTFB for at least 75% of the time.
• Mobile: 7 out of 17 Host have Fast/Average TTFB for at least 75% of the time.

Aggregate Statistics:

• Desktop: Fast (32.5%), Average (41.0%), Slow (26.5%)
• Mobile: Fast (32.2%), Average (46.7%), Slow (21.0%)

Page Volume Distribution:

• Desktop: Fast (11 out of 17), Average (10 out of 17), Slow (1 out of 17)
• Mobile: Fast (7 out of 17), Average (8 out of 17), Slow (9 out of 17)

6.2.11 Time to First Byte (TTFB) by Host, US

General Overview:

• Desktop: 14 out of 18 Host have Fast/Average TTFB for at least 75% of the time.
• Mobile: 12 out of 18 Host have Fast/Average TTFB for at least 75% of the time.

Aggregate Statistics:

• Desktop: Fast (40.6%), Average (43.0%), Slow (16.5%)
• Mobile: Fast (35.5%), Average (45.3%), Slow (19.2%)

Page Volume Distribution:

• Desktop: Fast (8 out of 18), Average (9 out of 18), Slow (11 out of 18)
• Mobile: Fast (8 out of 18), Average (7 out of 18), Slow (12 out of 18)

General Overview:

• Desktop: 12 out of 17 Host have Fast/Average TTFB for at least 75% of the time.
• Mobile: 9 out of 18 Host have Fast/Average TTFB for at least 75% of the time.

Aggregate Statistics:

• Desktop: Fast (36.3%), Average (41.7%), Slow (22.1%)
• Mobile: Fast (31.4%), Average (43.4%), Slow (25.1%)

Page Volume Distribution:

• Desktop: Fast (14 out of 17), Average (10 out of 17), Slow (3 out of 17)
• Mobile: Fast (11 out of 18), Average (12 out of 18), Slow (8 out of 18)

6.3 Time to First Byte (TTFB) by Content Delivery Network (CDN), Per Geographical Segment

6.3.1 Time to First Byte (TTFB) by CDN, Australia

6.3.2 Time to First Byte (TTFB) by CDN, Brazil

6.3.3 Time to First Byte (TTFB) by CDN, Canada

6.3.4 Time to First Byte (TTFB) by CDN, China

6.3.5 Time to First Byte (TTFB) by CDN, France

6.3.6 Time to First Byte (TTFB) by CDN, Germany

6.3.7 Time to First Byte (TTFB) by CDN, India

6.3.8 Time to First Byte (TTFB) by CDN, Japan

6.3.9 Time to First Byte (TTFB) by CDN, Russia

6.3.10 Time to First Byte (TTFB) by CDN, UK

6.3.11 Time to First Byte (TTFB) by CDN, US