{"id":4406,"date":"2021-06-10T22:08:57","date_gmt":"2021-06-10T22:08:57","guid":{"rendered":"https:\/\/www.waterscapetech.com\/?p=4406"},"modified":"2021-06-11T00:21:06","modified_gmt":"2021-06-11T00:21:06","slug":"we-analyzed-208k-webpages-heres-what-we-learned-about-core-web-vitals-and-ux","status":"publish","type":"post","link":"https:\/\/www.waterscapetech.com\/we-analyzed-208k-webpages-heres-what-we-learned-about-core-web-vitals-and-ux","title":{"rendered":"We Analyzed 208K Webpages. Here\u2019s What We Learned About Core Web Vitals and UX"},"content":{"rendered":"<div class=\"css-12m1dn8\">\n<p>We analyzed 208,085 webpages to learn more about Core Web Vitals.<\/p>\n<p>First, we established benchmarks for Cumulative Layout Shift, First Input Delay, and Largest Contentful Paint.<\/p>\n<p>Then, we looked into the correlation between Core Web Vitals and user experience metrics (like bounce rate).<\/p>\n<p>Thanks to data provided by\u00a0<a title=\"WebCEO\" href=\"https:\/\/www.webceo.com\/\" target=\"_blank\" rel=\"noopener\">WebCEO<\/a>, we were able to uncover some interesting findings.<\/p>\n<p>Let\u2019s dive right into the data.<\/p>\n<p class=\"d-block h2 text-center\">Here is a Summary of Our Key Findings:<\/p>\n<p>1.\u00a0<strong>53.77% of sites had a good Largest Contentful Paint (LCP) score.<\/strong>\u00a046.23% of sites had \u201cpoor\u201d or \u201cneeds improvement\u201d LCP ratings.<\/p>\n<p>2.\u00a0<strong>53.85% of websites in our data set had optimal First Input Delay (FID) ratings.<\/strong>\u00a0Only 8.57% of sites had a \u201cpoor\u201d FID score.<\/p>\n<p>3.\u00a0<strong>65.13% of analyzed sites boasted good optimal Cumulative Layout Shift (CLS) scores.<\/strong><\/p>\n<p>4. The average LCP of the sites we analyzed clocked in at\u00a0<strong>2,386 milliseconds<\/strong>.<\/p>\n<p>5. Average FID was\u00a0<strong>137.74 milliseconds<\/strong>.<\/p>\n<p>6. The mean CLS score was\u00a0<strong>0.14<\/strong>. This is slightly higher than the optimal score.<\/p>\n<p>7. The most common issues impacting LCP were\u00a0<strong>high request counts and large transfer sizes<\/strong>.<\/p>\n<p>8. Large layout shifts were the #1 cause of poor CLS scores.<\/p>\n<p>9. The most common issue affecting FID was\u00a0<strong>an inefficient cache policy<\/strong>.<\/p>\n<p>10. There was\u00a0<strong>a weak correlation between Core Web Vital scores and UX metrics<\/strong>.<\/p>\n<p>11. We did find that\u00a0<strong>FID did tend to slightly correlate with page views<\/strong>.<\/p>\n<h2 id=\"optimal-largest-contentful-paint-score\" class=\"text-center\">53.77% of Websites Had an Optimal Largest Contentful Paint Score<\/h2>\n<p>Our first goal was to see how each site performed based on\u00a0<a title=\"Google Core Web Vitals Report\" href=\"https:\/\/support.google.com\/webmasters\/answer\/9205520?hl=en\" target=\"_blank\" rel=\"noopener\">the three factors that make up Google\u2019s Core Web Vitals<\/a>: Largest Contentful Paint, Cumulative Layout Shift, and First Input Delay.<\/p>\n<p><img loading=\"lazy\" decoding=\"async\" class=\"alignnone size-full wp-image-4408\" src=\"https:\/\/www.waterscapetech.com\/wp-content\/uploads\/2021\/06\/all-core-web-vitals-are-part-of-googles-overall-evaluation-of-page-experience.jpg\" alt=\"core-web-vitals-are-part-of-googles-overall-evaluation-of-page-experience\" width=\"1528\" height=\"1528\" srcset=\"https:\/\/www.waterscapetech.com\/wp-content\/uploads\/2021\/06\/all-core-web-vitals-are-part-of-googles-overall-evaluation-of-page-experience.jpg 1528w, https:\/\/www.waterscapetech.com\/wp-content\/uploads\/2021\/06\/all-core-web-vitals-are-part-of-googles-overall-evaluation-of-page-experience-1280x1280.jpg 1280w, https:\/\/www.waterscapetech.com\/wp-content\/uploads\/2021\/06\/all-core-web-vitals-are-part-of-googles-overall-evaluation-of-page-experience-980x980.jpg 980w, https:\/\/www.waterscapetech.com\/wp-content\/uploads\/2021\/06\/all-core-web-vitals-are-part-of-googles-overall-evaluation-of-page-experience-480x480.jpg 480w\" sizes=\"(min-width: 0px) and (max-width: 480px) 480px, (min-width: 481px) and (max-width: 980px) 980px, (min-width: 981px) and (max-width: 1280px) 1280px, (min-width: 1281px) 1528px, 100vw\" \/><\/p>\n<\/div>\n<p>Specifically, we wanted to determine the percentage of pages that were classified as \u201cgood\u201d, \u201cneeds improvement\u201d and \u201cpoor\u201d inside of each site\u2019s Search Console.<\/p>\n<p>To do this, we analyzed anonymized Google Search Console data from 208k pages (approximately 20k total sites).<\/p>\n<p>Our first task: analyze\u00a0<a title=\"Large Contentful Paint\" href=\"https:\/\/web.dev\/lcp\/\" target=\"_blank\" rel=\"noopener\">LCP (Large Contentful Paint)<\/a>. In simple terms, LCP measures how long it takes a page to load its visible content.<\/p>\n<p>Here\u2019s how the sites that we analyzed fared:<\/p>\n<ul>\n<li>Good: 53.77%<\/li>\n<li>Needs Improvement: 28.76%<\/li>\n<li>Poor: 17.47%<\/li>\n<\/ul>\n<p>As you can see, the majority of sites that we looked at had a \u201cgood\u201d LCP rating. This was higher than expected, especially when taking into account other benchmarking efforts (like\u00a0<a title=\"Core Web Vitals Scores For Every Industry\" href=\"https:\/\/corewebvitals.iprospect.com\/\" target=\"_blank\" rel=\"noopener\">this one by iProspect<\/a>).<\/p>\n<p>It may be that the websites in our dataset are especially diligent about page performance. Or it may be partly due to a sample size difference (the iProspect analysis continuously monitors 1,500 sites. We analyzed 20,000+).<\/p>\n<p>Either way, it\u2019s encouraging to see that only about half of all websites need to work on their LCP.<\/p>\n<h2 id=\"first-input-delay-ratings\" class=\"text-center\">53.85% of Websites We Analyzed Had Good First Input Delay Ratings<\/h2>\n<p>Next, we looked at Search Console reported\u00a0<a title=\"First Input Delay\" href=\"https:\/\/web.dev\/fid\/\" target=\"_blank\" rel=\"noopener\">First Input Delay (FID)<\/a>\u00a0ratings. As the name suggests, FIP measures the delay between the first request and a user being able to input something (like typing in a username).<\/p>\n<p>Here\u2019s a breakdown of FID scores from our dataset:<\/p>\n<ul>\n<li>Good: 53.85%<\/li>\n<li>Needs Improvement: 37.58%<\/li>\n<li>Poor: 8.57%<\/li>\n<\/ul>\n<p>Again, just over half of the sites we looked at had \u201cgood\u201d FID ratings.<\/p>\n<p>Interestingly, very few (8.57%) had \u201cpoor\u201d scores. This shows that a relatively small number of sites are likely to be negatively affected once Google incorporates FID into their algorithm.<\/p>\n<h2 id=\"optimal-cumulative-layout-shift-score\" class=\"text-center\">65.13% of Sites Had an Optimal Cumulative Layout Shift Score<\/h2>\n<p>Finally, we looked at the\u00a0<a title=\"Cumulative Layout Shift\" href=\"https:\/\/web.dev\/cls\/\" target=\"_blank\" rel=\"noopener\">Cumulative Layout Shift (CLS)<\/a>\u00a0ratings from Search Console.<\/p>\n<p>CLS is a measurement of how elements on a page move while loading. Pages that are relatively stable through the loading process have high (good) CLS scores.<\/p>\n<p>Here were the CLS ratings among the sites that we analyzed:<\/p>\n<ul>\n<li>Good: 65.13%<\/li>\n<li>Needs Improvement: 17.03%<\/li>\n<li>Poor: 17.84%<\/li>\n<\/ul>\n<p>Among the three Core Web Vitals scores, CLS tended to be the least problematic. In fact, only around 35% of the sites that we analyzed need to work on their CLS.<\/p>\n<h2 id=\"average-lcp\" class=\"text-center\">Average LCP Is 2,836 Milliseconds<\/h2>\n<p>Next, we wanted to establish benchmarks for each Core Web Vital metric. As mentioned above, Google has created\u00a0<a title=\"Web Vitals\" href=\"https:\/\/web.dev\/vitals\/\" target=\"_blank\" rel=\"noopener\">their own set of guidelines<\/a>\u00a0for each Core Web Vital.<\/p>\n<p>(For example, a \u201cgood\u201d LCP is considered to be under 2.5 seconds.)<\/p>\n<p>However, we hadn\u2019t seen a large-scale analysis that attempted to benchmark each Core Web Vital metric \u201cin the wild\u201d.<\/p>\n<p>First, we benchmarked LCP scores for the sites in our database.<\/p>\n<p>Among the sites that we analyzed, the average LCP turned out to be 2,836 Milliseconds (2.8 seconds).<\/p>\n<p>Here were the most common issues that negatively impacted LCP performance:<\/p>\n<ul>\n<li>High request counts and large transfer sizes (100% of pages)<\/li>\n<li>High network round-trip time (100% of pages)<\/li>\n<li>Critical request chains (98.9% of pages)<\/li>\n<li>High initial server response time (57.4% of pages)<\/li>\n<li>Images not served in next-gen format (44.6% of pages)<\/li>\n<\/ul>\n<p>Overall, 100% of pages had high LCP scores at least partly due to \u201cHigh request counts and large transfer sizes\u201d. In other words, pages that are heavy with excess code, large file sizes, or both.<\/p>\n<p>This finding is in line with another analysis that we did which found that large pages tended to be the culprit behind most slow-loading pages.<\/p>\n<h2 id=\"average-fid\" class=\"text-center\">Average FID Is 137.4 Milliseconds<\/h2>\n<p>We then looked at FID scores among the pages in our dataset.<\/p>\n<p>Overall, the mean First Input Delay was 137.4 milliseconds:<\/p>\n<p>Here are the most prevalent FID-related issues that we discovered:<\/p>\n<ul>\n<li>Inefficient cache policy (87.4% of pages)<\/li>\n<li>Long main-thread tasks (78.4% of pages)<\/li>\n<li>Unused JavaScript (54.1% of pages)<\/li>\n<li>Unused CSS (38.7% of pages)<\/li>\n<li>Excessive Document Object Model size (22.3% of pages)<\/li>\n<\/ul>\n<p>It was interesting to see that caching issues tended to negatively affect FID more than any other problem. And, not surprisingly, poorly-optimized code (in the form of unused JS and CSS) was behind many high FID scores.<\/p>\n<h2 id=\"average-cls\" class=\"text-center\">Average CLS Is .14<\/h2>\n<p>We discovered that the average CLS score is .14.<\/p>\n<p>This metric specifically looks at how the content on a page \u201cshifts\u201d.Anything below .1 is rated as \u201cgood\u201d in Search Console.<\/p>\n<p>The most common issues affecting the projects\u2019 CLS included:<\/p>\n<ul>\n<li>Large layout shifts (94.5% of pages)<\/li>\n<li>Render-blocking resources (86.3% of pages)<\/li>\n<li>Text hidden during web font load (82.6% of pages)<\/li>\n<li>Not preloaded key requests (26.7% of pages)<\/li>\n<li>Improperly sized images (24.7% of pages)<\/li>\n<\/ul>\n<h2 id=\"how-lcp-correlates-with-user-behavior\" class=\"text-center\">How LCP Correlates With User Behavior<\/h2>\n<p>Now that benchmarks were set, we then set to find out how accurately Core Web Vitals represent real-life user experience.<\/p>\n<p>In fact, this relationship is something that Google themselves highlight in their \u201cCore Web Vitals report\u201d documentation:<\/p>\n<p>To analyze Core Web Vitals and their impact on UX, we decided to look at three UX metrics designed to represent user behavior on webpages:<\/p>\n<ul>\n<li>Bounce rate (% of users leaving a website\u2019s page upon visiting it)<\/li>\n<li>Page depth per session (how many pages users see before leaving the website)<\/li>\n<li>Time on website (how much time users spend on a website in a single session)<\/li>\n<\/ul>\n<p>Our hypothesis was as follows: if you improve a website\u2019s Core Web Vitals, it will positively affect UX metrics.<\/p>\n<p>In other words, a site with \u201cgood\u201d Core Web Vitals will have a lower bounce rate, longer sessions, and higher page views. Fortunately, in addition to Search Console data, this data set also contained UX metrics from Google Analytics.<\/p>\n<p>Then, we simply had to compare each website\u2019s Core Web Vitals against each UX metric. You can find our results for LCP below:<\/p>\n<p><a href=\"https:\/\/backlinko.com\/core-web-vitals-study\" target=\"_blank\" rel=\"noopener\"><strong>Read Full Article Here<\/strong><\/a><\/p>\n","protected":false},"excerpt":{"rendered":"<p>We analyzed 208,085 webpages to learn more about Core Web Vitals. First, we established benchmarks for Cumulative Layout Shift, First Input Delay, and Largest Contentful Paint. Then, we looked into the correlation between Core Web Vitals and user experience metrics (like bounce rate). Thanks to data provided by\u00a0WebCEO, we were able to uncover some interesting [&hellip;]<\/p>\n","protected":false},"author":3,"featured_media":4407,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_et_pb_use_builder":"","_et_pb_old_content":"","_et_gb_content_width":"","footnotes":""},"categories":[28,3,38,30,52],"tags":[115,116],"_links":{"self":[{"href":"https:\/\/www.waterscapetech.com\/wp-json\/wp\/v2\/posts\/4406"}],"collection":[{"href":"https:\/\/www.waterscapetech.com\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.waterscapetech.com\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.waterscapetech.com\/wp-json\/wp\/v2\/users\/3"}],"replies":[{"embeddable":true,"href":"https:\/\/www.waterscapetech.com\/wp-json\/wp\/v2\/comments?post=4406"}],"version-history":[{"count":0,"href":"https:\/\/www.waterscapetech.com\/wp-json\/wp\/v2\/posts\/4406\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.waterscapetech.com\/wp-json\/wp\/v2\/media\/4407"}],"wp:attachment":[{"href":"https:\/\/www.waterscapetech.com\/wp-json\/wp\/v2\/media?parent=4406"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.waterscapetech.com\/wp-json\/wp\/v2\/categories?post=4406"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.waterscapetech.com\/wp-json\/wp\/v2\/tags?post=4406"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}