Let's talk about AMP

Since its inception, Google's controversial web framework has divided opinion.

Nov 01, 2021

With the current kerfuffle around AMP as part of the broader lawsuit against Google, this is as good a time as any to talk about the divisive web framework.

I have thoroughly documented my own opinions on AMP previously, so I won’t reiterate the arguments I made there. I want to discuss something else that’s been grating me for several months now.

There’s this particular graph that, whenever I think about it - and what it actually means - it makes me angry. The more I think about that graph, the angrier I get.

This is the graph in question:

It shows the percentage of articles in Google’s mobile Top Stories carousel in the US that are not AMP articles. The sudden spike in non-AMP articles coincides with Google officially removing the AMP requirement for mobile Top Stories in the middle of July 2021.

Before then, non-AMP articles accounted for single-digit percentage of results shown in Top Stories on mobile devices. Afterwards, when any article - regardless of the technology it is built on - can rank in Top Stories, the percentage of non-AMP shot up to 25% for Google US (where it still sits today).

Let’s take a moment to digest what that actually means.

The AMP Offer You Couldn’t Refuse

Since its inception in 2015, AMP has been force-fed to publishers. The arrangement was simple: adopt AMP or forego most of your mobile search traffic. Without having valid AMP versions of your articles, your content would be extremely unlikely to be shown in Google’s mobile Top Stories carousels.

And that would hurt. A lot.

Back then, mobile use of the web was rapidly becoming the de-facto use case for most audiences, and (as we’ve discussed before) Top Stories carousels account for the vast majority of organic search traffic to publishers.

So there wasn’t really a choice. Adopt AMP or lose traffic. Lots and lots of traffic.

I’m not even going to talk about the inadequacies of AMP when it became the requirement for mobile Top Stories in 2016. Suffice to say that adopting AMP came at a huge cost for publishers, both in maintaining a parallel development track for AMP articles (with a rapidly evolving standard where the goalpost were constantly and unilaterally moved), and in limited monetisation opportunities for AMP (some of which are now the subject of litigation).

There’s a word for this kind of unbalanced relationship, where one party makes demands the other can choose to accept or suffer unpleasant consequences. It’s called blackmail.

Many publishers caved. Some didn’t. And that’s the crux of the matter.

The Cost of Non-AMP

Let’s look at that graph again:

That surge in non-AMP articles in Top Stories, that’s what makes me angry. From single-digit percentages of non-AMP to 25% non-AMP, almost overnight.

What does it really mean?

Superficially, it’s simple. Google no longer required publishers to use AMP for visibility in mobile Top Stories, so more non-AMP articles started to rank.

Articles that, previously, were ineligible for mobile Top Stories because they weren’t built on the AMP framework.

Articles that, otherwise, would have been shown. Articles that were valid and rankable in every way - in terms of content and quality and everything else Google considers - except for that one AMP requirement.

Undoubtedly, there is a lot of gratitude among publishers that decided to forego AMP and are now finally seeing their articles rank on mobile Top Stories. That surge to 25% non-AMP represents a huge amount of traffic.

Google handles trillions of searches every year, the majority of which are done on mobile devices. On average, a Top Stories carousel is shown on about 1 in 10 Google results pages. So we’re talking about hundreds of billions of search results pages where suddenly non-AMP articles are more abundant.

So, yes, these publishers are happy Google changed its requirement.

But I’m angry.

Because it means that for more than five long years, when AMP was a mobile Top Stories requirement, Google penalised these publishers for not using AMP.

There was no other reason for Google to stop ranking these publishers in their mobile Top Stories carousel. As is evident from the surge of non-AMP articles, there are likely hundreds - if not thousands - of publishers who ticked every single ranking box that Google demanded; quality news content, easily crawlable and indexable technology stack, good editorial authority signals, and so on.

But they didn’t use AMP. So Google didn’t rank them.

Think for a moment about the cost of that. How many visits these publishers didn’t get, simply because they didn’t accept Google’s blackmail. How much revenue these publishers lost because of that. How many jobs were affected. The compromises some have had to make just to survive. The ones that didn’t survive.

Just because Google demanded we embrace their pet AMP project.

It’s Not Google’s Traffic

I’ve heard the counterargument that publishers shouldn’t complain, as they had no right to that traffic. All those countless billions of lost visits, they weren’t ‘lost’ at all because the traffic didn’t belong to the publishers in the first place. These people visit Google, so Google can do with them what they want.

But that’s a very warped argument. Because those visits don’t belong to Google either.

People don’t go to Google to ‘read Google’. There is no Google-produced content there for the audience to consume.

Google was always the conduit, the mechanism through which an audience finds the content it wants to read.

To build on a very old analogy, on the digital highway of the internet Google served as the signposts that made sure you got to where you needed to go.

Over time, however, Google turned itself into a two-way tollbooth, extracting value from all parties connected to the highway.

And those who didn’t pay the Google toll (by not adopting AMP) were selectively disconnected from the highway (stopped showing in Top Stories).

The Future of AMP

There will continue to be some voices who promote AMP, positioning it as a very usable framework to build fast websites with good UX. And they’re right, AMP can certainly be used for that.

But for me, AMP is a toxic brand. Despite efforts to position it as nominally ‘open source’, AMP is still a Google pet project.

For me, that means it cannot be trusted. Google will continue to use AMP for its own purposes, and those purposes are often opposed to a free and open web - as is becoming increasingly evident.

Publishers are starting to walk away from AMP, and the early results are promising indeed.

For a long time I advised my clients to maintain AMP, lest they lose the bulk of their mobile search traffic. Now, I’ve changed my recommendation. Where feasible, publishers should experiment with removing AMP. Disable it for a section or two, and see what happens.

The resources that go into maintaining AMP pages should be redirected to improving the regular site’s core web vitals, and AMP should be retired in stages.

I genuinely hope that AMP will disappear, that it will be abandoned as a project.

It may find a future under a different name, hopefully with a different set of people at the helm who don’t work for Google in any capacity. In that guise, it may survive and even thrive as a framework for website development.

But the AMP brand is forever associated with Google’s blackmail, and the enormous cost that publishers have had to bear as a result.

I, for one, will happily dance on AMP’s grave.

How To Delete AMP

Getting rid of alternate AMP versions of your articles is relatively straightforward. You should complete the following steps in order:

1. Remove the ‘amphtml’ reference from your standard articles

To let Google know you have AMP articles, there is a link reference to the AMP version of an article in the regular article’s HTML source. It will look something like this:

<link rel="amphtml" href="https://www.example.com/article/amp" />

Remove this amphtml link reference from your regular articles’ HTML source code to let Google know you don’t have alternative AMP versions.

2. Redirect your AMP URLs

Next, you will want to redirect your existing AMP articles’ URLs to their standard version with a 301 permanent redirect. This makes sure that anyone (Googlebot and end users both) trying to access the AMP article will be sent to the regular version of the article.

So an AMP article on https://www.example.com/article/amp should 301-redirect to https://www.example.com/article.

3. Submit Updated XML Sitemaps

If you had included your AMP articles in your XML sitemaps, make sure your sitemaps no longer reference AMP URLs and then re-submit them to Google Search Console.

Since including AMP in XML sitemaps was always optional (and quite rare), this is likely not a necessary step in the process. But better to double-check and make sure your XML sitemaps don’t include any AMP links anymore.

4. Update the AMP Cache

To get your existing AMP articles out of Google’s AMP Cache, you can trigger a cache update via their API. The relevant documentation explains how to do this.

After having done all this, you should see a steady decline in indexed AMP articles in your Google Search Console report. Note that it’ll take time for Google to de-index your AMP articles, so it may take a while for the AMP graphs in GSC to reach zero.

Google has published official documentation on how to remove AMP, probably to pre-empt the deluge of questions they received since dropping the AMP requirement.

NESS Debrief

Last week the inaugural News and Editorial SEO Summit (NESS) was held. This online event was a specialised conference specifically for SEO for news publishers.

I helped co-organise the event together with John Shehata, and we’re both immensely proud to have put on this event. The initial feedback from attendees has been truly amazing.

The attendees came from all over the world to be part of the NESS, with more than 46 countries represented:

So far everyone who filled in the attendee feedback form has said they’d attend another NESS event, so we’ve already started planning the next edition.

All the talks of the first event have been recorded and can be viewed by ticket holders in the Airmeet platform for the next few days. We will be making all videos available for viewing through private YouTube links later this week, and will also send download links to the slides to all attendees.

For those who were unable to buy a ticket for the event in time, we’ll be launching a ‘video only’ ticket where you can buy access to the recorded videos of the talks. Keep an eye on the newsseo.io website for details.

Miscellanea

Some interesting bits and pieces from the past few weeks:

Google is reportedly working on a ‘Big Moments’ feature for breaking news, where it will use machine learning systems to highlight the most relevant content for developing news stories.

Testing headlines with A/B split tests is common among online publishers. However, finding a magic formula for successful headlines is proving difficult indeed.

The WTF is SEO newsletters continue to be must-reads. One of the latest editions discusses topical authority and how to improve it.

Bing has launched its live indexing API called ‘IndexNow’. This allows websites to pro-actively let Bing know when they have new & updated content to index.

You can now get Discover and Google News performance data out of Google Search Console via their API.

I can’t NOT mention the latest disclosures in the Texas vs Google lawsuit. This whole Twitter thread is worthwhile to digest, and this WSJ article summarises the main points.

That’s all for this edition of SEO for Google News. As always, thanks for reading and subscribing, and please share this with anyone you think may find it useful.