Since the announcement of the project in October 2015, a lot has been said about Accelerated Mobile Pages (AMP) — what they are, how to get started, how Google and other platforms use them and so on. Yet there is always more to learn.
With more than one hundred releases/updates so far, the AMP project is evolving rapidly, adding more and more supported features and acquiring more and more adopters.
At SMX Advanced last week in Seattle, Rudy Galfi, product manager at Google for the AMP Project, and John Shehata, VP of SEO at Condé Nast, presented exclusive information and data around AMP.
Galfi started the session by quickly covering the basics and explaining that the project was developed as way to make the mobile web faster and more user-friendly. The goals or requirements were well-defined: make pages fast, be easy to implement, enable monetization and embrace the open web. AMP, today, does all of this.
What is AMP?
Galfi took some time to expand on caching, as it is a “core key of how AMP works.” He explained that Google AMP cache functions in a similar manner as a content delivery network (CDN), that it is free for anyone to use and that it works on “stale-while-revalidate” model. This model helps make sure the content is always up to date in the cache. The process by which it works is quite simple: When a request is made, the client receives the cached version while the document is requested again from its original server to be updated in the cache.
AMP at Google
Galfi then shared interesting statistics from Google. The median load time for AMP is 0.7 seconds, the time it takes for your eye to blink twice. By contrast, the median load time for non-AMPs is 22.0 seconds, the time it takes for you to leave the site and never come back.
AMP launched in Google web search last February and continues to be deployed across Google platforms. The way Google is using AMP is also evolving, and Galfi’s team continually works on improving the user experience. For example, allowing users to swipe from one article to another in Google search appeared to be a key improvement and helped increase the number of articles read per session.
AMP adoption and statistics
With more and more users reading AMP articles in Google search, the integration of analytics, ads, content access and the release of plugins for WordPress and Drupal, the adaption metrics are now quite significant: 125M+ documents indexed, across 640K+ domains, according to Google.
John Shehata provided additional data from Searchmetrics showing the increase in adoption. From the analyzed websites, only three percent were using AMP in March 2016, versus 11.6 percent in June 2016.
Shehata also shared some interesting aggregate data from his clients: “For a strong news site, AMP is about 5% of total traffic (9% of total mobile traffic), 3% of total impressions, +2% in CTR and +5.6% in ranking position.”
Galfi mentioned that he’s not surprised about these numbers; having fast web pages has been a challenge for developers, and AMP helps to get there.
While AMP is mainly used for news articles and blog posts, Galfi explained that AMP is not restricted to this type of content. Pages that are “usable in static (cached) form and lazy loading friendly” are suited for AMP. This can be recipes, product listings, travel guides and so forth. AMP being applicable to several content types will also increase adoption over time.
How AMP works
The AMP publishing flow is pretty straightforward, but for it to work, a couple of things need to be done. The main requirement is that the page needs to be valid and discoverable. Additionally, if the page aims to be displayed in Google’s Top Stories carousel, it needs to be correctly marked up with structured data from Schema.org.
Validating AMP URLs
Individual URL validation
Another tool, mentioned by John Shehata and developed by me, allows you to test and edit your HTML code for validation. I am, however, retiring this tool, as Google and the AMP Project team just released a similar (but better) tool that you can find at https://validator.ampproject.org/.
For structured data markup validation, use Google’s Structured Data Testing Tool.
The other tool mentioned by Shehata is a bulk AMP validator on TechnicalSEO.com (Disclaimer: I also developed this tool). This one allows you to validate up to 100 URLs at a time, check their discoverability (canonical and amp-html tags), the presence of Schema.org markup and the potential link to the Google AMP cache.
Shehata also explained how Condé Nast was using a Slack bot to “relay messages from the CMS to Engineering when a newly published page does not validate.”
Google Search Console reports
Google, through the Search Console, also provides a validation report, giving you an overview of potential issues across your site. The report will give you information not only about the validity of your pages according the AMP specifications, but also about the validity of your Schema.org markup.
In addition, Google Search Console now returns AMP impressions and clicks through the Search Analytics reports.
Opportunities and challenges
Shehata, who worked on several AMP implementation projects with Condé Nast, shared valuable information about roadblocks his team had to face. He explained, for example, that while the WordPress plugin was helpful and saved a lot of development time, teams still have to work on customization to meet their needs. He also mentioned some issues around the design of pages and, more specifically, the integration of custom fonts.
Overall, the AMP project is a success; it evolves quickly and supports more and more features, and as a result, it gets more adopters. Both Galfi and Shehata said it: Developers love AMP, it’s relatively easy, and it fulfills a longtime need on the web — speed.
Also, remember than AMP is not restricted to news publishers. We can assume that AMP URLs will, sooner than later, be displayed in Google for several types of content. And, as is often in the search world, early adopters will win.
Check out Rudy Galfi and John Shehata’s presentations below:
Opinions expressed in this article are those of the guest author and not necessarily Search Engine Land. Staff authors are listed here.