but the Mobile pages are still about 55%.

This is normal, in that a typical mobile view simply takes the desktop content and squeezes it into one very tall vertical layout.

This leads to two extremes, on one hand the prime content typically covers one screen length, so the user reads the content and leaves the page without going any further. At the opposite extreme there is lot of content and the user get annoyed with scrolling down and ditches the page early. In both case ads lower down on the page are loaded but never seen. Also, ads at the top of the page may not be seen either if they are slow to load. This is results because the user will immediately start scrolling down before seeing the ad appears.

It is strange that, on desktop, you saw a drop and subsequent recovery after switching from responsive ads. Are you sure you fixed the issue you posted about previously where the responsive ads weren't appearing? Did you fix the issue on both the desktop and mobile layouts? Did you fix it on all ad units?

Hi @NickMNS,

Thank you very much for your nice answer. I'm afraid that this weird figures come from the fact that, at the same time of including responsive units and merging the contents, I removed the AMP pages for mobile.

It seems that AMP pages load ads in a smarter way, which I'm not able to replicate in my non-AMP mobile pages. And it seems that the Desktop users find easier to scroll down up to the end of the pages, with the consequent higher 'Active View Viewable' figures.

I recently worked on improve CLS and page load times, after much testing I see that revenue and ads shown is down around 20%.

Traffic is constant, and to give an idea of the sample I'm working from, the daily traffic unique visits is around 7,000 people with 45% UK 45% US and 10% rest of world. Server is UK based, and add revenue for the UK is much lower than the US, (at around 20%), presumably because the site is faster in the UK and also the EU cookie regulations have dented ad revenue.

For reference all ads are automatically placed, and I have one flexible template for all devices that scales and changes div layouts to fit the screen.

The page load time was reduced from 4.5 seconds, to 2.5 seconds. (HTTP2, CDN tweaks, Minify, cutting code down/removing, streamlining javascript, removing junk code from Wordpress, and compressing images.)

Ads at the top of the page are not showing any more. I've concluded that if a page loads too fast a use will scroll down and less ads will be shown. I feel annoyed that I spend so much time following recommendations only to suffer a drop in income.

I reverted back to the old site theme and revenue popped right back to where it was expected again with ads appearing at the top. So I'm not going to bother with page load speed, and will just focus on CLS which didn't seem to impact ad display or revenue.

Also look at your visitor metrics, particularly bounce rate and time on site, this can indicate a problem with the new layout.

I've found that preloading the main ad script helps lift revenue, and requesting ads as high up in the page code as possible also helps.

I'm going to try slowing up the page load times for certain regions and devices and see what that does. Interestingly by loading a totally pointless jquery migrate library, additional javascript libraries and 5 unused fonts. Page load times are back up to around 4 seconds and revenue and ads shown shot up. This needs a little fine tuning but it is interesting.

The most important speed mod one can do is improving TTFB and getting your page code downloaded as quickly as possible, the next priority is the ad code, then load images and iframes etc last.

In my experience after a site change, google ad servers need a short while to adjust to the changes, give it a week, measure, change then another week and measure again.

Hi @waynne,

Thank you very much for sharing your experience. It's a very educative post.

Ads at the top of the page are not showing any more. I've concluded that if a page loads too fast a use will scroll down and less ads will be shown. I feel annoyed that I spend so much time following recommendations only to suffer a drop in income.

I share that experience. I managed to reduce the page load time to less than 1 second (as you did, optimizing JavaScript, minify, leaving the less important elements to be loaded after scrolling, etc.). And yes, ads at the top have a lower AVV rate (and incomes) than the rest of the ads. However, I'm still measuring, and I don't dare to remove that top ads, which was historically the best units.

So I'm not going to bother with page load speed, and will just focus on CLS which didn't seem to impact ad display or revenue.

I understand that you won't bother with pagespaeed anymore after having reduced it from 4.5 to 2.5 seconds, right? I agree also with that. I think that CLS leads to a poorer user experience (and Google penalties) rather than the pagespeed. I started to use responsive units (because Google was suggesting them to me), but I started to suffer CLS issues and removed them.

I started to use again fixed-size units, and now I'm experimenting with the "data-full-width-responsive=true" for mobile users. Do these full-width mobile ads create also CLS issues?

I've found that preloading the main ad script helps lift revenue

What do you mean exactly? To use the following piece of code?

<link rel="preload" href="https://pagead2.googlesyndication.com/pagead/js/adsbygoogle.js" as="script">

I'm using this method to load the AdSense JavaScript library. Would it work with it?

var el = document.createElement('script');
e1.async = "async", el.src = 'https://pagead2.googlesyndication.com/pagead/js/adsbygoogle.js';
var node = document.getElementsByTagName('script')[0];
node.parentNode.insertBefore(el, node);

requesting ads as high up in the page code as possible also helps.

I do not get this point. Do yo mean to execute the 'push' command as high up in the HTML code?

I'm going to try slowing up the page load times for certain regions and devices and see what that does.

Please share the conclusions of this experiment. Anyway, will not Google SERPs penalize this if you start to show different JavaScript code to slow down the pageload?

In my experience after a site change, google ad servers need a short while to adjust to the changes, give it a week, measure, change then another week and measure again.

Yep, this is one lesson I've learnt about all this. Let Google time to learn ad adjust to the changes while you measure. Do not try to change the layout of the ads every day.