AMP is responsible for 43% of search traffic in 2017. This is driving a major uptick in sessions for the AMP pages.
Overall search traffic has risen substantially over the last year, which in part is due to fully embracing AMP. They have seen growth of nearly 70% YoY in search traffic, with AMP garnering nearly 50% of that growth.
They have no perception of any rankings increase (other than being in the news carousel).
Roughly 31% of SERP ranks are from AMP, based on data in July 2017. These changeovers to AMP happened pretty automatically. In terms of the traffic gain, they had little effect until they were accepted into Google News in June 2016.
Page views per session in AMP are typically only about two-thirds of what they are in the normal mobile experience. They are also seeing a lower bounce rate and higher session duration for non-AMP. However, with the swipe design of AMP and the lack of our usual related modules and infinite scroll, they don’t think that the comparison is apples to apples, as their AMP pages have a very different UX.
As noted above, Thrillist sells the ads themselves, so there were no issues with third party ad platforms and their AMP readiness (e.g. do they support https).
They did not see any major issues with analytics (they use Google Analytics).
While they have seen tremendous growth in users due to AMP, they would love to see this product cover more content types in the future (e.g. better coverage for video).
The caching rules around AMP are not entirely transparent. They are not sure how long it takes for updates to reflect on Google servers. They have seen multiple instances where updates don’t show up in a timely manner.
They also see delays in updates in Search Console for AMP pages.
Mimicking their production environment was another concern. It's more difficult because AMP pages are served from a Google subdomain. As a result, they noticed a lot of CORS-related errors. (Cross-Origin Resource Sharing, or CORS, is an HTML5 feature that allows one site to access another site’s resources despite being under different domain names.)
They also had parts of articles that did not map 1:1, and issues with images because there were so many restrictions. These related mostly to how image data was delivered from their CSS. It was not easy to populate the AMP component, as they did not naturally have the image size (which is required by the AMP-img tag). They also had issues with embeds, videos, or anything that requires specific sizes to be specified.
They had to completely eliminate interactive elements, such as their sign-up form, but they believe that they may be able to do that now. They were able to implement forms in AMP on another site.
Generally, they had issues with content that is not supported by AMP. As a result, they implemented guidelines to prevent this from happening, and also cleaned some of these things up either manually or programmatically. They may have tweaked some ad sizes, how they work with advertisers, but no big deal.
They are very satisfied with ROI because primary KPI is driving session growth, and they’ve seen AMP traffic increase significantly over the last year.