AMP: Do or die? Session recap from SMX West
The “AMP: Do or Die?” session at SMX West covered the hot topic of Accelerated Mobile Pages (AMP).
AMP is the Google-backed project designed to speed up webpages on mobile devices. Google has actively been pushing adoption of AMP, but due to the many challenges inherent in migrating to AMP, adoption has been slow.
This session provided three contrasting perspectives of AMP. Moderated by Michelle Robbins of Third Door Media, the session had a split format which started with three formal presentations followed by a lively sit-down discussion where Google representative Cheney Tsai joined the panel in taking questions from the audience.
The first speaker was AMP evangelist Dave McAnaly of Resolution Media.
Dave provided a presentation entitled, “AMP: Being an Evangelist and a Realist,” in which he shared his agency’s perspective on AMP.
He began his talk by recapping the three core value propositions of AMP:
- To help content to load faster.
- To reduce abandonment.
- To cut down on the “clutter” on webpages.
Dave explained his agency’s clients who moved to AMP had experienced a positive lift with AMP, so they view it as a viable business solution. Improvements in visitor engagement with AMP content and mobile have made his company a proponent of moving to AMP.
Dave balanced his enthusiasm for AMP by describing some of the challenges associated with AMP that his company has experienced. Since AMP is a new format with a non-trivial learning curve and extra management and resource requirements, implementation can be expensive and take time. He also emphasized that AMP is not a replacement for user experience (UX) enhancements.
A big takeaway from Dave’s presentation was that AMP was not a “one and done” solution. It requires continual management, and it covers the scope of many competing stakeholders in a large company.
Dave recommended the easiest way to get people to embrace AMP is to use the “test and invest” approach. This involves picking a specific brand challenge (such as slow loading pages or pages that don’t perform well on mobile), then implementing AMP in that area. After letting the test run for a while, examine the results with real measurable data.
He described a case study where they did such a test with a client site. They had noticed other competitors with AMP pages were showing up high in the search results pages. They selected four pages from the client’s site needing improvement. After they moved the client’s pages to AMP, they gathered data for a few months and found they had achieved better than 2x faster load times and a 2x to 3x improvement in time on site for these pages.
Remarkably, the four pages they put in AMP produced a 10 percent lift in organic traffic on mobile devices.
Dave’s recommendation when moving to AMP is to not move an entire site into AMP at once. Instead, Dave encouraged taking baby steps and doing only a few pages. Then quantify and measure the changes. He emphasized that next steps need to be actionable and aligned with business priorities.
The second speaker was Pete Dainty, the Global Head of SEO at eBay.
His presentation, entitled “Lessons Learned Implementing AMP on eBay,” provided a realistic view of some of the difficulties encountered and the level of commitment it takes to succeed with AMP on a large site.
Pete explained that eBay was one of the first e-commerce sites to implement AMP. eBay started implementation in July 2016 and was fully live by October 2016. They partnered with the AMP Project Team to work through issues and they continue to update and improve their AMP adoption.
Pete summarized the pros and cons that eBay experienced with moving to AMP. He confirmed that tracking and attribution were harder with AMP, and moving to AMP required customization — it wasn’t an out-of-the-box template solution.
Site speed is critical to eBay’s success. Pete stated that AMP dramatically improved site speed for visitors coming from Google on mobile. After AMP, the Time To First Byte (TTFB) improved 44 percent, and the full load time improved 68 percent.
Pete noted that being able to use Google’s content delivery network (CDN) was extremely valuable — especially considering eBay’s size. While they could use another CDN, with 1.2 billion listings, migrating could be very expensive.
One of the takeaways from his presentation was that “AMP is essentially a free CDN.”
Pete identified a few of the problems his company encountered moving to AMP. He explained that when you move to AMP, you remain in Google’s ecosystem, so Google continues to control the UX elements. Pete stated, “For mobile-first indexes, AMP may not be able to show all the responsive web content from your core site.”
Pete also cautioned that AMP can eat your crawl budget and introduce complexity and new errors that will need addressing. He warned that attribution was harder with AMP. Conversion and bounce rates are difficult to capture, and other items like unique visitors can be inflated. Pete explained how new sessions can be generated between AMP and non-AMP which might inflate bounce and lower conversion rates. Also, AMP can sometimes be attributed to direct traffic, not Search. Pete said it will require customizing in most analytics packages.
At the end of his talk, Pete asked, “Is AMP worth it?” His response was that despite the issues with tracking and attribution and the extra custom work needed to resolve the problems, it was worth the move. Being able to leverage Google’s pre-rendering and free caching made the pages faster, thus improving the visitor experience. These benefits made adoption of AMP worthwhile for eBay.
The third speaker was Patrick Stox, technical SEO at IBM, who provided a contrasting (and less enthusiastic) viewpoint on AMP.
Describing himself as the “angry old man” of the panel, Patrick’s presentation explored why AMP may not be a good solution for all companies. Implementing AMP may be too difficult and carry an excessive risk for a large enterprise company with a complex structure.
Patrick explained that while the goal of AMP is to improve the web experience, often there is missing functionality. There may be a lot of business reasons for having certain items on your site that may not be able to be included in AMP.
He identified a number of “carrots” Google is using to encourage adoption. Some of the inducements he mentioned are the AMP Carousel, AMP stories, the Lightning Bolt icon and starting the website load from Google search engine results pages SERPS (HTTP Preload).
This preloading makes the loading of AMP pages look instant and gives AMP pages a real usability benefit. Patrick says without these incentives, no one would go through the headaches of moving to AMP.
He said that despite the advantages, there is pushback to AMP. As an example, Patrick mentioned AMPletter.org, a site where a group of developers have complained about AMP.
Working for a large enterprise company has extra pressures that make AMP adoption challenging. There are standards, guidelines and legal and business requirements that can make moving to AMP extra difficult. Combine that with the added complexity of yet another system to maintain, especially when it will need to be implemented across multiple systems and require buy-in from different teams who may or may not have the resources, and it makes moving to AMP nearly impossible for a large enterprise.
Patrick finished his discussion with a strong recommendation: If you are going to move to AMP, hire more developers, because you will need them.
The remainder of the session was a question-and-answer period with questions from the audience. In addition to the panelists, Cheney Tsai, senior solutions engineer at Google, joined the discussion.
One of the questions asked was (paraphrased) “Do you need to change to AMP, or are there enough benefits from just improving your site to make it load faster?”
Pete Dainty acknowledged that you could (and should) do a lot of enhancements outside of AMP to improve your site but that AMP puts an extra layer of benefits on top of your efforts. The pre-rendering, for example, is something you can’t get anywhere else.
Cheney Tsai added that AMP was a tool to make the web better for users. There may be other tools out there that create awesome user experiences and make responsive sites. AMP may not be as effective in a large enterprise environment. At the end of the day, AMP is just a tool to assist in attaining the goal of improving the web for users.
Another question was whether fast-loading non-AMP sites could be given the lightning bolt icon like AMP sites get on the search results pages. If the goal of the carrot (the visual lightning bolt) is as an incentive for site owners to increase the speed of a site, why not let all fast sites get the lightning bolt?
Cheney responded by saying that while he can’t speak for the Google Search team, there is a guarantee that an AMP site will be fast, so it makes sense to have the icon.
He reminded the audience that speed is already a ranking factor, so a fast site does get some benefits for being fast, even without the lightning bolt icon in the SERPS.
Google may not have the load times of all the sites in the SERPS, so it would be very difficult to predict a site’s performance. AMP is a guarantee of a speedy experience, so they are willing to tag AMP pages with the lightning bolt icon.