Forum Moderators: martinibuster
I have sites/pages with very different eCPMs. Earnings used to be ok until smart pricing hit hard on me recently. With not much to lose, I decided to do something about it. One way, as suggested by some, is to remove AdSense from pages that generate low earnings, but how do we identify such pages? A page may have high EPC (therefore high eCPM), but because of low impressions, it generates low earnings. Other pages may have low eCPM, but give decent income. I went with low eCPMs.
After removing AdSense from pages with extremely low eCPMs (< $1), overall earnings started to rise. But when I restored AdSense on these pages, "smart pricing" hit again. Seems to me that Google hates low eCPMs, so much so that it smart-prices entire accounts of publishers.
Does anyone have similar experiences?
Added: yes, we can run long term tests but over time so much changes on the net that it's hard to tell if the changes were because of something we did or something else.
For what it's worth, I've experimented with and without AdSense ads on my low-eCPM (but decent-EPC) photo galleries over the past couple of years, and I've never been able to discern any benefit from pulling the code. As with most things, different people may have different experiences, so testing is the best way to find out whether pulling AdSense code from low-eCPM pages will improve or hurt total AdSense revenues.
Some of us have done the testing, and found that (for us) removing ads from some pages where they don't work actually does increase revenue. However, we still have to endure the endless lectures and Googleworshipping that are stifling the debate that others want to have.
I don't class some minor tinkering some time back on a couple of pages as a meaningful test, and certainly not enough to enable pointification on the topic.
Oh - and just a little friendly tip here. All the best4 and top8's I see are probaly harming your bottom line. The recent tanking might have something to do with that.
On my photo galleries, for example, I tend to have low eCPMs from CPC ads because users tend to look at a lot of photos without clicking
EFV,
I realise that this may not necessarily suit the format of your particular site, but I have found that in my case I have achieved a good ctr with my photo gallery by having it launch in a small separate (borderless) window. This way when the user finishes looking at the images they refer back to the original window.
I have found that in my case I have achieved a good ctr with my photo gallery by having it launch in a small separate (borderless) window. This way when the user finishes looking at the images they refer back to the original window.
You mean, like a JavaScript window? Does it include ads? Are the photo-gallery pages indexed by search engines (as conventional HTML pages are)?
IMO. All we can do is run tests and see what works today, but might not work tomorrow or next month...
True, but testing is still a better way to find out what works or doesn't work for your site than relying on anecdotal evidence from people whose sites may be a lot different from yours.
This difference of standpoints makes it hard to properly differentiate the data since we don't currently have enough information to make a decision. IMO. All we can do is run tests and see what works today, but might not work tomorrow or next month.
This is a HUGE point. I go a little farther. You can run tests, and find out whether your revenue and numbers go up or down (and that's obviously important), but running tests doesn't tell you anything about how google works.
So, what works on one site may not on another.
What works on one set of pages may not work on another set.
Too many unknown interacting variables. One thing I've NEVER seen discussed here is the algorithms for how google chooses WHAT to display on a particular page. That's critical in understanding generally whether removal of adsense from some pages will work or not.
It's a complete black box.
If, for example, you remove some adblocks and that affects what ads are displayed in your remaining blocks in some unknown way, and you don't know that, you'll derive bad conclusions.
I've pretty much concluded after several years of this that while some tweaking is worthwhile, there's a point where it's better to stop worrying about google mind-reading, and focus back on visitors.
It's funny watching some of the sites owned by people here and elsewhere over time, because you can see the "experimentation" -- adding blocks, removing blocks, shifting to YPN, removing YPN, adding affiliate links, removing them, basically going back and forth eternally.
Which I guess is ok until you notice that during all the time, not a single piece of new content is added, and the visitor experience is ignored, or in fact gets worse with tweaking. Those are also the website owners that complain in forums like this, because it's clear they can't tweak their ways into sustainable revenue increases.
It's kind of like walmart changing the colors of their pricing tags every day or so, while selling crappy merchandize.
One thing I've NEVER seen discussed here is the algorithms for how google chooses WHAT to display on a particular page. That's critical in understanding generally whether removal of adsense from some pages will work or not.
Good point. From my own experience, I'd say it's fairly obvious that the criteria for selecting ads change from time to time, and that Google sometimes experiments with those criteria. If you made changes to your site based on the day when Google was serving ads for hamsters on your home page about cockatiels, you might be shooting yourself in the foot.
Some of us have done the testing, and found that (for us) removing ads from some pages where they don't work actually does increase revenue. However, we still have to endure the endless lectures and Googleworshipping that are stifling the debate that others want to have.
The key words in the statement above are for us. In other words, you've done what I'm suggesting, so why would you object to others being encouraged to the same thing? And why dirty the discussion with ad hominem references to "Googleworshipping" or "stifling the debate" just because some members may not agree with your conclusions? How does that contribute to the body of knowledge on the topic?
I don't class some minor tinkering some time back on a couple of pages as a meaningful test
I don't, either, which is why I've tested with more than 1,000 pages on several different occasions. And I've come to much the same conclusion as rbacal has done: The target is constantly moving, so--once you've made a decision based on testing--the best strategy may be to sit tight and focus on the site's content instead of constantly playing with ad code.
I've pretty much concluded after several years of this that while some tweaking is worthwhile, there's a point where it's better to stop worrying about google mind-reading, and focus back on visitors.
I agree with this. In fact I suspect a fair number of publishers have "tweaked" themselves right out of a good thing.
...adding blocks, removing blocks, shifting to YPN, removing YPN, adding affiliate links, removing them, basically going back and forth eternally.Which I guess is ok until you notice that during all the time, not a single piece of new content is added, and the visitor experience is ignored, ...
I'd be a little cautious with this one. On my site I like to tweak in areas that haven't been changed for a while, because they haven't been changed for a while.
That way I have a better chance of knowing the tweaking was responsiblle for any results in stead of other recent changes being responsible.
But that doesn't mean the whole site has been unchanged and a casual observer might not see all the new stuff if they don't look at all the site sections.
I'd be a little cautious with this one. On my site I like to tweak in areas that haven't been changed for a while, because they haven't been changed for a while.That way I have a better chance of knowing the tweaking was responsiblle for any results in stead of other recent changes being responsible.
But that doesn't mean the whole site has been unchanged and a casual observer might not see all the new stuff if they don't look at all the site sections.
I wasn't implying that tweaking excludes content development across the board, but that some people fall into the trap of trying to tweak and squeeze the last drops of income from their content, and they become so obsessed with it they forget about the reason why people might come to their sites, or stop developing their content and value.
For me personally, and in a business sense, and thinking long term, I need to ask myself where to put my time for long term value. Should I experiment with colors? Position? Number of blocks? adlinks? Removing blocks? Removing from some pages?
...or should I (once I've gotten to a point where nibbling at the details is pointless) look to making my site unique, valuable, linkable, trafficable, etc. and where the investment will pay off long term (but perhaps not short term), but it will pay off literally for years?
If the google system as an entirety was more predictable and knowlable, I'd think a bit differently. Too many variables. Too many unknowns, and trial and error is a really poor way to run a business, unless you don't know any better.
testing is still a better way to find out what works or doesn't work for your site than relying on anecdotal evidence from people whose sites may be a lot different from yours.
anecdotal evidence:
Based on personal observation, case study reports, or random investigations rather than systematic scientific evaluation: anecdotal evidence.
I'm not sure which evidence I'd rely on first. Trying a 50/50 split right away wouldn't really work since the results I am looking for are overall numbers. I'm leaning towards trying one method for 6 months then the other for 6 months. This way, next year, I will have one additional inexplicable piece of data to add into my equations, sounds like fun.
There was not a week/month in these past 3 months that have not seen my earnings go up, CTR and eCPM both are up about 80% at this point of the test.
Are total earnings up? Not just eCPM? That's what matters to the bottom line. (It's easy to boost numbers like EPC, CTR, and eCPM by removing ads from lower-performing pages, but the real test is whether total earnings improve.)
My earnings are almost double of what they were for the pre test period.
As I said before I encourage every adsense publisher to try/test this, but doing this for 2 days or a week is not a testing. You need a longer period and a lot of studying.
My earnings are almost double of what they were for the pre test period.
OK. Thanks.
As I said before I encourage every adsense publisher to try/test this, but doing this for 2 days or a week is not a testing. You need a longer period and a lot of studying.
Ideally the publisher would do random "split-run" testing, to minimize the number of other variables that could give false results (such as week-to-week variations in supply and demand, changes in Google algorithms, etc.). That wouldn't be practical on a static HTML site like mine, but I assume that a programming-savvy publisher who's using a CMS could figure out a way to do it.
[edited by: Broadway at 3:40 am (utc) on Oct. 16, 2006]
The key words in the statement above are for us. In other words, you've done what I'm suggesting, so why would you object to others being encouraged to the same thing? And why dirty the discussion with ad hominem references to "Googleworshipping" or "stifling the debate" just because some members may not agree with your conclusions? How does that contribute to the body of knowledge on the topic?
Some 30% of the posts in this thread are from you (about average for the course), and apart from fighting with Danimal (who made some very valid points), a lecture about ecpm being more useful than it is in this situation, dismissing the evidence of 4 people in the thread who have done what you haven't (achieved success in removing underperforming ad blocks), and a fair amount of general lecturing you haven't really made any attempt to answer the OP's question. Constantly dragging the subject off course onto what you want to talk about is stifling debate.
Googleworship - objectively read a few threads and you'll see. Everyone else can.
As regards doing what you are suggesting, I think you are very late onto the bandwagon and are trying to make it your own. In posting about this over the last year or more, I've always suggested experimenting and pointed out that although it's worked for me it may not work for others. I've never suggested otherwise. Unlike you, I don't expect compliance with my views - I point out what has worked for me, and try to keep to the point the OP actually asked in the first place.
The target is constantly moving, so--once you've made a decision based on testing--the best strategy may be to sit tight and focus on the site's content instead of constantly playing with ad code.
I don't constantly changing layouts either. I never have. My main earning pages have been virtually unchanged for some time now. Yes, like most people here I do experiment from time to time, but my rule is that it it's working well, then leave it alone!
However, to try and drag back the debate to what the OP was discussing (what pages to remove adsense from) my strategy is this:-
Look at ALL the metrics available, If a page is getting a much higher number of impressions than other pages do and the ecpm is low, then that may be a candidate for removal and requires further monitoring or analysis. It's possible that the disproportionate number of loads without clicks may affect smartpricing in some way. I'm not necessarily suggesting that ctr is the key, or ecpm is either. As someone said earlier, it's a black box and you know no more on the workings than I do. If a page has a low ecpm and low number of visitors, then that page is unlikely to affect overall site earnings and would probably stay. A lot of low earning pages can add up to a decent overall income.
But the overall point is that if ads don't work on a page, or an ad block doesnt work then there is no reason to keep it. It's not earning you money and may be hampering performance of othe ad blocks. It's certainly adding to ad blindness, and that's the main reason I don't show ads on every single page on my site.
You mean, like a JavaScript window? Does it include ads? Are the photo-gallery pages indexed by search engines (as conventional HTML pages are)?
Yes I use a javascript window, however it does not contain ads. The launching page contains the links and a description of each gallery along with small icon images and a single adsense skyscraper. The photo gallery windows are not indexed but the gallery page is and functions as the 'hub' of my website.
[edited by: Scurramunga at 8:54 am (utc) on Oct. 16, 2006]
David_uk wrote:But the overall point is that if ads don't work on a page, or an ad block doesnt work then there is no reason to keep it. It's not earning you money and may be hampering performance of othe ad blocks. It's certainly adding to ad blindness, and that's the main reason I don't show ads on every single page on my site.
eCPM = CTR x EPC x 10
About practical details, I think the point is not low eCPM, but very low eCPM, that is to say non-performing ads. Since those are non-earning ads, even when they are on high traffic pages, we can lose, say, less than 5% of earnings the first days after the removal. But one or two weeks later the total earnings often go up.
Of course, there is a great variety of results, and this cleaning seems to work for some sites and not for others. However, if we are just interested in the user experience, we do it anyway.
[edited by: Juan_G at 10:13 am (utc) on Oct. 16, 2006]
Sometimes, especially if we improve those non-performing pages or make them more useful and topic-focused, they get better ads and start to perform.
[edited by: Juan_G at 10:49 am (utc) on Oct. 16, 2006]
Some 30% of the posts in this thread are from you (about average for the course), and apart from fighting with Danimal (who made some very valid points), a lecture about ecpm being more useful than it is in this situation, dismissing the evidence of 4 people in the thread who have done what you haven't (achieved success in removing underperforming ad blocks), and a fair amount of general lecturing you haven't really made any attempt to answer the OP's question. Constantly dragging the subject off course onto what you want to talk about is stifling debate.
EFV and I disagree on a lot of things ;) but I'm surprised that the obvious common sense in his eCPM assertions need explanation and defence! Feel free to not take up any suggestion made here, or elsewhere. I've been exerting my own freedom to studiously avoid all this anecdotal rubbish masquerading as, what do you call it, "evidence"? Four separate sources for this evidence must make it infallible ;) but I'll continue ignoring it on the way to new record Adsense earnings almost every month.
[edited by: Juan_G at 11:34 am (utc) on Oct. 17, 2006]
At least for one site I maintain, I've felt that CPM ads were counter productive, very low paid impressions, versus Pay per Click ads.
So on August 18th I politely asked Google to turn off CPM ads. By August 19th they had done so, and politely responded CPM ads could be restored at anytime, so this is a completely harmless change and test.
For periods of 10, 30, and almost 60 days before and after this change, EARNINGS after Aug 18th are consistently UP approximately 40%.
This was a step increase in earnings after turning off CPM ads. The site was not showing too many CPM ads and of course the performance for any given page was typically poor from a CPM stand point. I even looked at earnings earlier in the year when the site typically does better, but earnings are still higher after August 18th.
CPM ads may be a significant factor in this discussion and resulting analysis and conclusions. Frankly it's as if this jump in earnings returned the site to "pre smart pricing" days, but perhaps it was only to pre CPM ad days; that would take more research to determine.
If anyone has the date CPM ads were introduced I could look back further.
It could be some of the experimenters in this thread are actually turning off pages with CPM ads, and somehow, for a given site perhaps CPM ads impact "smart pricing" negatively. For reference this small site has about 700-900 page views/day.
So this perhaps is a tidbit I shouldn't spread around, but what the heck. I'd prefer this tip not be a new thread. Of course it could all be coincidence.
If anyone has the date CPM ads were introduced I could look back further.
April, 2005, according to this article [onedegree.ca].
Personally, I don't think site-targeting affected my site, but then I've never been site-targeted, as far as I know.
