I recently switched back to 100% / 100% settings because I saw dip in adsense revenue and I wasn't using any backup ads. The dip wasn't huge but it wasn't just 1% either.

I'm now showing 30% ads 70% blank space. The real question is are we seeing a better user experience?

2/3 less ads, pages may be loading up to 70% faster, visitors aren't seeing the cheap flashy ads, those are the ones bring filtered, user experience must be at an all time high. Is anyone seeing an improvement in Google Analytics? Pages per visit, avg time on site, bounce rate, site speed. All these metrics should show improvements.

Donald, that's been my main concern, too. I've found over the years that increased the load time increases the pages per session, which increases the overall revenue, so I expected the same thing.

It really hasn't, though. I started using the Ad Balancer on 1/28/17. On 1/27/17, I had an average page per session of 7.65, and on 1/28/17 it was 7.99.

But then a month later, on 2/24/17, the average was 7.56. Then on 2/25/17 (comparing Saturday to the 1/28/17 Saturday) the average was 7.98.

So I'm seeing virtually no change on pages per session. I broke it down to mobile, desktop, and tablet, too, and so virtually no change on any device.

I'm currently at 49% Revenue / 99% Coverage.

A much improved user experience would be if the ad unit (not being served) would collapse. Having a huge white space is hardly an improvement. Yes, I realize you can use a backup ad, but what if you don't want to.

I just checked my Analytics going back to when I first implemented the Ad Slider in mid January and was a bit disappointed to only see between 2% - 3.5% week over week improvement in pages per visitor, time on site and bounce rate. Numbers were slightly different on mobile vs desktop,

It's surprising because ads are always the major causes of issues with the various page speed tools I've used over the years, to the point that I can only get useful data by testing a page with no ads or 3rd party widgets (e.g. AddThis). Although, Adsense is by far the best performer compared to other Ad Networks I use where 1 ad will routinely make 50-100 extra requests.

And I just put myself into a state of shock. According to the Ad Slider 1% of my ads generate 45% of total revenue! That is just mind boggling and I'm tempted to test it.

Very good point, anefarious1. I wonder if you can use Javascript to manipulate the iframe from the backup ad source code. You would a backup "ad" that removes the parent iframe from the page. Should be possible if the backup ad page is on your domain, but I'm not sure how the iframe is loaded exactly, e.g. whether or not it's an iframe within an iframe. This would be easy to test if only there was a way to test backup ads. This thread [webmasterworld.com] does include some ideas, but I haven't tried them.

Donald, I just compared my page load time on Analytics, too.

Note that I began using the Ad Balancer on 1/27/17:

1/27/17, 6.97s
1/28/17, 6.42s
1/29/17, 7.49s
2/24/17, 7.33s
2/25/17, 7.95s

So there has been no apparent improvement in average load time, even after I went down to 49%. If anything, it's marginally worse!

My guess is that Adsense is going to run through everything and take the same time to decide whether to show something or not. And then you have to allow a little more time for the backup to load, so I guess it makes sense for the load time to suffer a little.

Sorry if the question is silly, but, does your load time stats at Analytics take in consideration the fact that ads are loaded asynchronously ?

@theDonald123

And I just put myself into a state of shock. According to the Ad Slider 1% of my ads generate 45% of total revenue!

There is no need to be shocked. Ad earnings per click vary greatly, very few ads pay a lot and many ads pay a little. This is reflected in the ad-balancer and you can see the distribution in the revenue profile report.

That is just mind boggling and I'm tempted to test it.

I do not recommend testing it. There is high variability and a large component of luck involved in terms of matching high paying ads with users willing to click them. By reducing the number of trials (each ad impression is a trial in this case) you will reduce the probability of reaching your full earnings potential. Simply put, if you reduce your impression too much, it will negatively impact your earnings. How much is too much, that is difficult to say and depends on a wide array of constantly changing factors. I would not go any lower than Adsense's "better UX" range but that is not based on anything other than my gut feeling.

@Nick

I've been treating the ad balancer like the floor price which other Ad Networks offer to publishers. For example in AdX every impression is an real time (RTB) auction. So, if Advertiser Y is willing to pay the most for this impression at $0.09 and your floor price is $0.10 then AdX does not display an ad. Setting a floor price can sometimes increase the winning bid in a real time ad auction, I think because it acts like an additional bid and increases the pressure. But it also reduces all the low quality inventory and low paying ads which contribute very little to total revenue.

You've explained it, probably correctly, that the ad balancer is Adsense's estimation that over a period of time displaying say 50% of impressions will result in the same total revenue . Adsense does not use RTB so maybe you are correct. Although there are also retargeting ads, not all Adsense ads pay per click.

I would think there would be more documentation for such a major new feature. Is Google able to guess which user is a high probability clicker and only show them ads? In PPC, publishers only make money from clicks so at most you are only making money from the ~3% who click. Maybe only 3% of internet users click on ads and Google knows who they are!

@theDonald123 this whole process is insanely complex.

So, if Advertiser Y is willing to pay the most for this impression at $0.09 and your floor price is $0.10 then AdX does not display an ad.

With a price floor, inventory remains opens for those willing to pay. Two scenarios can occur:
1 - If the demand at the price point of the floor is not sufficient to use up the supply the publisher will be left with un-filled ad slots. If a new ad buyer enters the market and over bids, then all that occurs is the regular player simply pick up the unsold inventory. If no-one clicks on these new ads, this new advertiser will have no impact.

2- If all the supply is consumed, and there is excess demand, this will push up the winning bid price for all the ads displayed, not just the new player. And then the impact will be felt regardless of whether the new high priced ad is filled or not.

A symptom of excess supply is the appearance of low quality ads. To remedy this you can set a floor or simply remove excess supply from the market. When you remove the excess supply you force the second scenario, I think that this is Adsense intention with this new feature.

The problem with this is that if you restrict supply too much, then you push prices up and advertisers may look elsewhere for a cheaper solution. Then in a worst case scenario you (publisher) are left with a bunch of advertisers with deep pockets and crap products. So they get all the impressions and you get no clicks. So maybe you would have been better of with scenario 1.

Then add to this, time of year, time of day, user mix, traffic fluctuations, macro economic conditions, and whole whack of other factors that will make your earnings vary from day to day, it is almost impossible to choose an optimal strategy.

Is Google able to guess which user is a high probability clicker and only show them ads?

Maybe to some extent, but I doubt that really makes such a big difference, they key is not only who but also when. Timing is everything. Will the user click on the ad on your site, or the next site they visit?

@Dimitri - could you please post a real example of backup ad html code? Or send one in PM? Thanks

@NickMNS - my mistake, you are right, Impressions are around half of what they've been ( logical to 57% Fill Rate)

So far I'm getting close to my previous earnings with 57% Fill Rate. If I can get a good aff offer or CPA, it could benefit me, unfortunately the niche in what I am is not very good with such offers.

>> @Dimitri - could you please post a real example of backup ad html code? Or send one in PM? Thanks <<

I am not sure what is you're problem but here is an example :

<!DOCTYPE html>
<html>

<head>
<meta charset="utf-8">
<style>body{margin:0px;padding:0px;}</style>
<meta name="robots" content="noarchive,noindex,nofollow">
<base target="_top">
</head>

<body>
<a href="xxxxx"><img src="xxxxx"></a>
</body>

</html>

nothing particular.

@CSdude, put the widget code in between the body and /body tags in a separate HTML file and save it to your site in an HTML file. Then enter the HTML file as a backup ad for an Adsense unit.

Is there a solution for testing this? I found out Adsense only allows an url of 256 characters but my url was to long. I can simply upload a html-file with the url placed into it and than use the url of the html-file in Adsense?

Edit: is it like the one here above? But never noticed tags like:

<head>
<meta charset="utf-8">
<style>body{margin:0px;padding:0px;}</style>
<meta name="robots" content="noarchive,noindex,nofollow">
<base target="_top">
</head>

<body>
<a href="xxxxx"><img src="xxxxx"></a>
</body>

</html>


I don't have to name a location of an image only a html text for another network or a longer than 256 characters url.

[edited by: Runfun at 1:14 pm (utc) on Mar 1, 2017]

@Runfun, are you sure you have an "URL" longer than 256 characters ?!

I'm not very sure, the example is:

<a href="http://---.com/click/click?p=1&t=url&s=---&url=https%3A//---.com/nl/elektronica/index.html&f=BAN&name=Audio &amp; beeld" target="_blank"><img src="---.com/nl/upload/banners/soundvision-17412_pp_728x90.jpg" width="728" height="90" alt="Audio &amp; beeld" /></a><img src="---.com/click/impression?p=1&s=---&t=url&f=BAN&name=Audio &amp; beeld" width="1" height="1" />

I've changed few parts of the code or was it not even possible to use this link?

Your post at 8:57am is right, put it in an HTML file like that and then link to the HTML file.

Like this?

<!DOCTYPE html>
<html>

<head>
<meta charset="utf-8">
<style>body{margin:0px;padding:0px;}</style>
<meta name="robots" content="noarchive,noindex,nofollow">
<base target="_top">
</head>

<body>
CODE AS I PLACED ABOVE
</body>

</html>

Thanks if this would be the solution.

@Dimitri - thanks, will try it

One more thing about the Ad Balance. After testing with 99% Revenue/57% Fill rate until the end of February, that's almost two weeks it shows that Adsense has deducted around 2,6% from my earnings for February.
It is unusually high, as I'm getting no more than 1,5-1,8%. I wanted to reach the end of month to see if for showing less ads I will get less invalid clicks to deduct at the end of the month.
Today I reverted to 100% Revenue/80% Fill rate and will keep it until the end of March to see if that makes any change.

How does Adsense determine what the fill rate should be to maintain 100% earnings? How accurate will this be for sites with relatively low traffic? My account is showing 100% rev at 30% fill. What is the time period adsense looks back on to gauge this balance. I have noticed my cpc increased quite dramatically in the last few days when I set the fill rate at a pretty low level, but that could also be because my site is coming out of a sustained period in which adsense performed poorly (based on historic performance). So I am not sure what to attribute this to.

How does Adsense determine what the fill rate should be to maintain 100% earnings?

They do not disclose how they determine the values of the ad-balance feature. I find this somewhat frustrating. But as far as I can tell it is based on the distribution of ads that appeared on you site over relatively short time period (day or two). Look at the "revenue profile" report in Adsense. What the system does is takes that curve and chops off the left side. How far left depends on the setting.

How accurate will this be for sites with relatively low traffic?

Accuracy has more to do with the slope of the revenue profile curve, a flat curve will generate some wild results because little shifts up or down can move the target (left, right position) by big margins. Say the change in the slope between 90% and 25% is ($0.07 @50%. - $0.05 @90%) = $0.02. Now say you set the ad balance at %50, meaning you forgo any ads that will pay you less than $0.07 (on average, it is set to % not $ value). Then suddenly the market changes slightly and moves up by a penny so %50 is at 0.08. But the slope stays the same, now you are suddenly forgoing 20% more ads that initially anticipated. Since now 70% of your ads are earning $0.07 or more, but you are only showing 50% of them due to your initial setting.

I don't know if this clear but the basic idea is:
If you have a flat curve over most of the range then the ad-balance is probably not a good tool for you site. The ad-balance tool really works well when you have a nicely sloped line that flattens at the left end. In this case, the ad-balancer is preventing what essentially is your un-sellable ad space to be removed from the market. Conversely, if there is no market at all for you ad space (the whole of your curve is flat), then the tool will basically block everything. In which case you may want to look at your website targeting, in attempt to find some interested advertisers.

My account is showing 100% rev at 30% fill.

See my comment above

What is the time period adsense looks back on to gauge this balance.

Not clear, as far as I can tell it is a few days at most.

I have noticed my cpc increased quite dramatically in the last few days when I set the fill rate at a pretty low level, but that could also be because my site is coming out of a sustained period in which adsense performed poorly (based on historic performance). So I am not sure what to attribute this to.

It is expected that your CPC increases dramatically, The CPC is the average CPC over time period selected by you in the report. If you remove all the ad clicks that paid you only a penny or two, then one naturally expect the average to go up. Example: before you had 10 click per day, 9 x $0.03, 1 x $0.73 == avg CPC of $0.10. If the you block impression and clicks for ads paying less than $0.05 per click, you are left with 1 click at $0.73, so your CPC for the same period is now $0.73. But you only earned $0.73 instead of $1.00. This does not mean that you are earning more money, it is only a side effect of the reporting. Similarly CTR and RPM should also go up, because you have fewer impressions, again this an expected reporting side-effect and will have not impact on your earnings. IMO what you observed has little to do with poor past performance.

With all that being said, the biggest benefit for me, has been the ability of ad-balance to have noticeable positive impact on eliminating spammy ads. I do not have any hard evidence or proof, because so many other things have been changing over the same period that I have been using the ad-balance, but I feel that the elimination of the spam ads has provided a better experience for my users that allows them to see only quality ads, this in turn must have a positive impact on earnings.

Nick, My CPC has not only increased these last few days after shifting the ad balance metre, the overrall earnings in each of those days were signficantly better than daily earnings for many months. I too have noticed lot of spammy ads getting through to the point I don't exclude them on individual basis, I am hoping the ad balance would do this. But I can't help feel 30 percent simply isn't enough ad views to garner a good rate of clickthroughs on a consistent basis. Also, considering how poorly adsense has perfomed for me for sometime now, the fill rate is likely to be very inaccurate. Isn't it better for me to run at 100% for a while (assuming this higher cpc is sustained) to provide the ad balance tool a more realistic benchmark.

Yes, I pushed the slider down to about 38% which was the top of the best user experience range with 100% of earnings. Well it wasn't 100%. My income dipped by about 25% and the backup affiliate ads I put in place didn't make up for it. Now I'm at 58% and my income shot up.

I don't trust the numbers it shows; I knocked mine down - keeping 100% revenue - and have seen a 25% decrease for the past few weeks. I've pushed it back up again.

In any case, AMP ads don't collapse, so I have big white gaps in the page. I'm not sure that's a better user experience really; the only potentially good user experience is that it speeds up the page load, but I've not seen that either...

(double post)

Every time I adjust the ad balance and go back a day or two later it keeps lowering the "Better user experience range" threshold. In other words, it always wants me to keep dropping the # of ads that I show.

I currently have it set to 57% ads showing with 100% revenue, but if I drop it down to the range, it will be at 54% ads with 99% revenue. I've already lowered it down 2-3 times, each time not dropping it below 99%.

So should I keep dropping it down every few days to keep it in the better UX range or just stop fudging with it? I'm not seeing an impact on earnings, although April has been overall down so it's really hard to do a fair comparison. At the same time I'm trying to figure out the best way to monetize the unused inventory (it's huge).

My experience is that if you leave it alone it stabilizes somewhat. The better UX moves up and down a little but mostly it stays in the same range. I have my balancer set at 60%, better UX is around 50%. I have sneaky suspicion that the better UX simply marks the 99% revenue mark.

My ad balance had been set at 50% for some time and it was showing 100% income and better UX just under 50%. I have started moving the balance up and had it set at 75% for the last week. Today the better UX experience was at 67% and I was surprised to see that it was showing income at 99% and not 100.
It appears to me that the better UX has little value, G just puts it somewhere below your current balance setting.

Of course the variations we see could be due to the ad pool itself changing over time - possibly seasonal changes in my case. I also believe that you can learn from this and adjust your ad numbers and positions. If you improve your ad placements, then surely you can increase the ad balance to benefit from the improvements you have made.

This is certainly not a set and forget tool.

As an aside to this - while I had the slider on "80% of impressions, 100% of revenue", I moved it back up to 100% a few days ago and, surprise surprise, I'm seeing about 20% more revenue...

There is something I do not trust in the ad balance tool, which is why I am not (yet) daring to use it.

I understand the idea, Google knows that xx % of your ads are generation most of the incomes, and so by displaying only these ads, this should be the same.

But, that is based on the previous days / weeks performances of these ads. But what about "new" ads which are constantly being added ? When are they evaluated and how? How Google determines if a new ad will be in the top earners ?

Also, does the ad balance works only for contextual ads? Or is it also applying to the personalized ads ?