Why the bounce rate of AMP pages is so high
christianoliveira.com
christianoliveira.com
I also in general find AMP pages to be a bit of a UX nightmare.
Outside of raw speed, I have nothing good to say about AMP pages, and the speed is nowhere near good enough to justify using them. They make the Web a worse place for the sole benefit of Google. They are abusing their monopoly position in search by forcing AMP pages down people's throats.
In fact, my bounce rate is high on AMP because I can near-instantly figure out if this particular article on this topic I searched is the one I actually want to read or if it's uninformative trash. I usually flip through 2-5 before settling on the one I think is good.
These are the same sorts of complaints people have made about every change to search ranking. And while, yeah, I'm sure Google's not being altruistic, the implications of most of their actions in these spaces is pretty consistent:
If you want people's attention, don't ship trash.
BUT, the problem stated in the article has nothing to do with that behaviour. It refers to the inability to track the user journey correctly due to Google's AMP distribution system. It doesn't matter if you click on an AMP result an go back in 2 seconds, or if you click on an AMP result, read 30 seconds, and click on a link inside the content that takes you to the original website: both cases are considered bounces.
Just like they also rob you of forcing me to take a survey first, or hounding me for my email, or otherwise preventing me from doing the single thing I want to do (read your content).
Your site should be sticky because you have good content, not because you've tricked me into clicking more links.
Is this opt-in for the publishers?
How different is this from from browser's blocking pop ups?
I prefer not to using a pop up blocker, so I experience the web like the general population. But I was so sick of publishers forcing ad clicks on mobile by effectively blocking scrolling on smaller form factor devices like my iPhone SE, that I've switched to the Brave browser.
2. Google did and still does. Load times were still unacceptable across the board, usually not because they had to be but often times because the people running the websites don't care enough to. Same as for HTTPS.
3. AMP is a more drastic reaction. It appears to be working.
That being said, I would like to see AMP become enacted only on sites that do actually take a long while to load. At the same time, AMP is still opt in. This isn't forced altruism, but you can play nicely or be lowered in rankings in response.
-- edited for formatting
Personally I'd prefer that Google not play party whip in setting behavior patterns; I'd rather it be as decentralized as possible. To me that's one of the advantages of the web over the app ecosystem: power is more atomized.
I'm a firmly bitter user who has stopped frequenting a lot of websites because they don't load well or don't give me what I consider to be an acceptable value prop (I'll happily turn off my adblocker if your ads are well behaved, I won't take a survey to read a one paragraph article).
But most users, and especially most non-technical users, aren't willing to take that same stance despite being very frustrated. It seems to vary between "well that's just how the internet is" to "I don't know any other sources". Having google prioritize non-terrible sources is a big boon to those non-technical users in my eyes, cemented firmly when I watch my mom or other older family members try to search the internet.
That's where you can help! The next time you visit your parents or non-techie friends, install something like Ghostery for them. It's not a long-term solution to web awfulness, but it makes the web less terrible for now.
I got so tired of sites with scraped content behind ads and logins. It's not perfect but it helps.
Why? Because users aren't power users. Most don't know better.
I admit I do keep track of a handful of bad offenders in my head, but I really shouldn't have to do that. Sure, it would be better if there was a trusted decentralized system to enforce rules and guarantee a decent UX, perhaps with more granularity so people could set different acceptable points, but until that exists, of course I'm going to use the presence of an instant articles or AMP icon to guide my browsing.
But how exactly are you suggesting I browse the web? Not following links unless I recognize the destination and know it to be a reasonable site? That just changes a mental blacklist into a whitelist. And how would I ever discover new reasonable sites?
Use an ad blocker? That feels like escalating a arms race, and I'd like to avoid that if at all possible. (I currently use Disconnect, which is only a tracking blocker, not an ad blocker, although some sites seem to think it is.)
Yes, I'd also recommend a JS blocker. If you don't want to do that, that's OK, but don't expect any sympathy if you get compromised in any way. If you don't lock your front door, don't be surprised if someone breaks into your house.
If a site obstructs the content in any way, or starts playing sound and I didn't expect it (for example video sites may be acceptable) I hit back immiedately.
I don't expect most users behave in this way.
I do agree that if Google really want to do anything to help the cause, they would more heavily lower the ranking (or completely remove from their index) any sites that don't meet basic performance and readability thresholds.
AMP is clearly a play into their own walled garden.
This seems like a much tougher problem than AMP.
From an API design standpoint, it is much easier to build and maintain an endpoint that receives data that must be in a specific, proprietary data format instead of an endpoint that allows the user to input data in any format and then attempts to parse that data after receiving it.
What is the problem here?
Sometimes I want the source article because it had more features (comments, related articles, etc), and sometimes AMP is just glitchy.
The fact that people here on HN are still requesting it says it's far from well implemented - I wouldn't have known about it if I hadn't read it somewhere when you added it. The more conspiratorial among us probably assume that implementing it poorly is intentional.
Is it because "Sharing" in this case us meant to imply pushing the content to another medium (such as an external browser) vs. just the implied meaning of "sharing with another person?"
I'd love any context you can share from internal discussions around how prominent to make that functionality. I think it is hard to deny that Google benefits from it being harder to find, which is where the criticism from the parent is likely coming from.
Also: using browser's native sharing option, shares the Google URL's, not the "real" one
I literally loaded Chrome for Android after reading your comment, searched for a Reddit thread, and no button, nor any other way to bypass AMP without having clicked the AMP link already.
Screenshot: http://i.imgur.com/Xnrlzcs.png
Before, there was no way to get the original URL. Now they display a "link" icon on the top so you can get the original URL.
From the results themselves, no, there is no way to bypass AMP (like there is with App-indexing)
I find messing with the user agent (mobile firefox plugin "phony" lets you pose as lynx, for example!) can be effective, and also I saw on a google products discussion page that using the url https://news.google.com/news/i#0 returns non-AMP results.
Why it can't just be a user option to disable it, I've no idea.
Google amp could play an important role in the future if net neutrality changes significantly. It also brings up some questions about whose content you're actually reading if Google is acting as a middle man between producer and consumer. In the worst case scenario, the amp'd version of a page could be different from what the original author intended.
> In the AMP ecosystem, the platform that links to content may freely choose which AMP Cache (if any) to use. It is an inversion of the typical model where content delivery is the responsibility of the publisher.
For the end user, AMP links are clearly marked, and non-AMP news results are readily available.
For the content producer, Google only publishes their content via AMP if the content producer permits it. Indeed, it is up to the content producer to make the AMP version of their content.
Content producers are being strong-armed into this by Google because AMP capable pages are being given preferential treatment in the SERPS.
They also:
- break the web
- hijack content
- reduce the value for the publisher
- make it impossible for the publisher to monetize their content
- are a step back from an open web
Pretty soon you'll be 'certified information provider' to Google rather than web publisher.Why stop at mobile?
Okay, how? AMP is built on HTML5 standards.
>hijack content
AMP Cache is different than AMP. Do CDNs also "hijack content", by serving them from faster servers?
>reduce the value for the publisher
By offering them an option to make pages load faster?
>make it impossible for the publisher to monetize their content
Incorrect. Read up on the allowed tags in AMP. Both ads and analytics are possible; they're just delayed until after the content is shown.
>are a step back from an open web
Built on open web standards.
> Okay, how? AMP is built on HTML5 standards.
Will only comment on this: it should be easy to figure out a way to break the open web using HTML5 standards:
- browser sniffing
- crazy javascript navigation
etc. Just from the top of my head.
The AMP team is working with analytics vendors to display data as realistic as possible.
At AMPConf on Tuesday eBay gave a great talk how to deal with this phenomenon. Video will be out soon.
How anyone at google ever signed off on this is beyond me.
Also many sites are slow because of ads - and Google runs the biggest and slowest ad network of all with DoubleClick. AMP is probably the most convoluted solution for all of this, likely created as an experiment by a small internal team but then expanded by management that should've known better. Now sites have to spend limited dev resources on maintaining a proprietary fork of HTML just for google traffic rather than working on better universal sites. After all the noise about optimizations and responsive design, this is a big step back.
(Edit: not that this is necessarily great for publishers or the ecosystem!)
For me, I understood it as a HTML subset, a method for image scaling, and an overall goal of having clutter free sites that work well on mobile. I knew there was a cache somewhere, too. But that users stay on googles site and the url won't be shown was not clear to me.
There was a very popular article on here about someone who implemented AMP and got surprised by this, the header change of copying the original url goes back to this.
Opting out of caching requires making your page intentionally invalid, and thus you lose all of the presentation and pre-rendering benefits of AMP.
By now I read all the horror stories, have a better picture of what it technically actually is, and opted not to support it for my sites.
We measured "active time on page" based on how engaged the user was on the page: were they scrolling, clicking, moving the mouse on a focused page, etc. It wasn't perfect, but it was much more helpful to publishers than Google Analytics' page-level metrics.
My point is that bounce rate is meaningless, regardless of if you're looking at AMP or non-AMP content. Maybe that was stated?
The problem is not about bounce rate, is about cookies and the inability of tracking users correctly. Give it another read ;)
Is it just that you're not allowed to use third-party plugins for ads and tracking, and instead have to use a single standard Google plugin?
Similarly, page analytics are all collected once, by a single AMP library, and then distribuetd to as many endpoints as you want, instead of each analytics endpoint having to duplicate the work of collection.
There are trade-offs. AMP is restrictive (no third-party JavaScript, limits on CSS, etc.), the AMP library itself is 200 KB (60 KB gzipped), and AMP is ultimately controlled entirely by Google, though the core team is interested in recruiting external maintainers.
I'm still hopeful that publishers will bring a class action lawsuit against Google and force AMP's closure. Sooner the better.
I do still use google on desktop though, because it does help that google knows who I am to give me more relevant results (not that I agree that this should be how it works)
This was in response to FB doing Instant Articles which had much the same effect on publishers. They couldn't afford NOT to be on FB, but still lose in the end because publishers have zero leverage these days thanks to Aggregation Theory [1].
I mostly Google specific things or click links from HN.
or put another way, has anyone mistakenly under-reported their traffic numbers?