PageSpeed Insights is an online Google tool that Marfeel uses to measure the performance of a mobile page and verify Marfeel's industry-leading pagespeed. Marfeel uses it as a constant audit to ensure publishing partners always score in the top-third of the fastest mobile sites on the web delivered by implementing the most performance-enhancing engineering techniques.
From a user standpoint, this is an important resource because it evaluates a mobile site's pagespeed by quantifying the First Meaningful Paint (FMP) and DOM Content Loaded (DCL). It also assesses web development optimization by identifying if the page uses performance best practices.
From an SEO perspective, a publisher's PageSpeed score has become even more important because as of July 2018, Google uses pagespeed is an SEO ranking signal for mobile searches; therefore a good score here will result in better ranking for publishers.
Getting a good score in PageSpeed Insights is an exceptional challenge considering that Marfeel is measured taking advertisements into consideration. Being at the pinnacle of all assessments with advertisement and a full set of features taken into account is what Marfeel strives for, and the cornerstone for all successful publishers on the mobile web.
The data PageSpeed Insights uses
For the speed sections, PageSpeed Insights uses data from the Chrome User Experience Report (UXR) to produce its results. The report is a public data set of UX metrics for top web origins that are aggregated from Chrome users that opt-in to synchronizing their browsing history. Therefore the results for speed are from real-world interactions and browsing that a publisher's users have experienced when accessing their publication.
If a test is executed for a URL not in the Chrome UXR, the speed score will be unnavailable and users will see the following message:
For the speed assessments, the test's data set contains metrics for the previous 30 days which are updated weekly. For optimization scores, PageSpeed caches results for a short period but should recognize any changes within minutes.
What PageSpeed Insights measures
Marfeel applies 100% of the optimizations that can be developed, however sometimes, the test doesn't produce a score of 100 for some Marfeelized pages. In the majority of cases, this is due to advertisers following a non-optimal cache policy for certain assets due to the dynamicism of their business.
The best practices a publisher is already using are represented in the Optimization Suggestions section. The optimization best practices that still need to be implemented are displayed in the Optimization Suggestions section.
Possible results vary between Good, Medium, and Low. A score is determined by how many of the defined best practices still need to be implemented. The more best practices missing, the lower the score.
An acceptable score here is "Good."
The DCL is important because it measures how long it takes for the HTML document to be loaded and parsed. Marfeel achieves industry-leading DCL times because there's no dependency on third-parties, assets are lazy loaded, and page elements are rendered as late as possible and pre-fetched as soon as possible.
Page Load Distributions
The Page Load Distributions section displays the different experiences users have had regarding the FCP and DCL when accessing a publisher's web page, according to the metrics collected by the Chrome User Experience Report. There are three possible scores; "Fast" (the top third results aggregated from Chrome's report), "Average" (middle third), and "Slow" (bottom third).
This section and metrics give publishers a real-world understanding of the browsing experience their audience encounters; it provides a breakdown of how a publisher's audience expriences the publication's pagespeed (taken from the Chrome's data set), implying the different connection speeds, connectivity quality, etc. an audience encounters.
This section assesses the additional round trips that need to be made to the server to load the page's render-blocking resources and the total bytes required to load the page.
PageSpeed Insights then provides the average roundtrips and total bytes needed to load a page across the web according to its data set. As in the example, Marfeel requires 0 additional roundtrips to load render blocking resources.
Demos and partners that have not activated yet
As explained in the Data PageSpeed Insights uses section, if partners test their Marfeel demo or version that is yet to go live (by entering the corresponding Marfeel domain where their Marfeelized content is hosted - b.marfeelcache.com, bc.marfeelcache.com, or enjoy.marfeelcahce.com), the test will only yield an optimization score and there will be no speed score because the URL is not in the Chrome UXR data set.
For newly activated publishers, PageSpeed Insights uses metrics from Chrome's UXR from the previous 30 days and updates weekly, so it will take some time before Marfeel's impact is reflected. Optimizations however are cached and recognized by the test promptly.
Marfeel is committed to keeping publishers' PageSpeed scores at a consistently high level so they can reep the SEO benefits that come with it. To deliver this, Marfeel vigilantly monitors the PageSpeed scores for all Marfeel publishing partners and holistically tracks this in our internal monitoring and reporting tools.
Specifically, PageSpeed test scores are tracked and recorded once a day and accumulated in a time series database. Automatic alarms are also configured to automatically go off if there are any trends or changes in the scores' behavior. This proactively alerts Marfeel support teams to review why the score has changed, and if any regression has been introduced or not.