Forum Moderators: martinibuster
This strategy won’t increase the potential earnings of your site – only additional content or functionality will do that – but it will bring your site’s earnings nearer to its potential. That is, it won’t increase your site ceiling, but it will take your earnings more consistently to that ceiling.
Background
I had significant success last year with an EPC-based strategy. After experimenting with various other approaches, I was prompted to return to this strategy by a recent thread [webmasterworld.com] started by Chris999. This proved to be an important piece in the jigsaw puzzle of smartpricing. Since then the pattern of last year has been repeated: earnings have been at a consistently high level, compared with much greater ups and downs previously. In fact, average daily earnings are now running at a record level, despite the slight decline in visitor numbers, normal for my business-oriented site as we approach Easter.
Why focus on EPC?
EPC, indirectly, provides “customer feedback”. Whilst Smartprice has an account-wide element, updated once or possibly twice a week, it also has a page (URL-oriented) element that seems to be updated daily. Although Google don’t provide us with data on conversions and other smartpricing factors, EPC gives us feedback on the impact of smartprice.
If, when measured over a month’s timescale, EPC on a page is in the same ballpark as Adwords prices, then the page is probably not being smartpriced downwards, or not by much. But if the page EPC is significantly lower, then the smartprice algorithm reckons that the page is a problem for some reason, eg: poor conversion rate or content that is not directly relevant to advertisers. That is, EPC (when related to Adwords rates) provides quantifiable customer-feedback on the performance of the page.
As with other businesses, the key to success is to identify where there is poor customer feedback, identify the cause of that poor feedback, and then resolve the problem. When I’ve taken this EPC-based strategy on my site, the benefits have been an increase in site income, junk ads disappearing off the site (removing the need to block them), and an improvement in site stickiness.
There is no guarantee that the strategy works for you, but I lay it out here so that, if you have the time and inclination, you can give it a try. The strategy also has some other spin-off benefits, such as better quality clicks for advertisers, increased income for Google, less income for purveyors of junk ads, and if lots more people did it, a better reputation for the Adsense scheme.
The EPC-based strategy
The first step is to collect relevant information. You need to know:
The second step is to try to improve EPC on the worst-performing pages. You can do this by following other advice in this forum, eg: moving ads to above the fold, reducing the number of ads on the page, ensuring that the most visible ad block loads first, etc..
The third step is to monitor your changes at least until a couple of days after your next account-wide smartprice update (I’ve described how to work out when that takes place in my Golden Gate post [webmasterworld.com]).
The fourth and final step is to remove ads altogether from low-EPC pages. Replace the ads with links to other pages on your site, promoting higher EPC pages. Don’t try to dupe users into going to pages to click on ads, try to maximise the chances that they will find a page relevant to them.
Final comments
This strategy depends on you having a content-rich site, with easy site navigation, and your willingness to spend time optimising your most popular pages. Although it requires an investment of time, it is mostly a one-off investment and will repay your efforts.
Some people talk about loading higher paying ads first.
I track most ads by channel. On one of my higher earning pages, the link unit has significantly lower earnings and CTR than the center content large square.
However, the link unit is first in code of execution on page. Is there a way to have the large square in -content ads load first?
Also, I don't see it, but is there a way to limit the number of ads Google shows on dynamic pages with most than one post? In other words, a way to only allow Google to show one ad on a page with many posts that are created dynamically from a database.
Thanks for all of your help.
Cainer
A few other questions, as it really seems you know alot about this.
Sadly, these particular questions are not my forte. In fact, we don't use link units, and our page design is rather simple by comparison with what you are trying to do. As soon as someone starts talking about dynamic pages I give them the phone number of the technical guy who does our dynamic stuff. I'm more concerned with the business side of the website.
Perhaps someone else can answer CainIV's questions above?
I'm defining 'low performing' in a very specific way (by comparing average EPC for the last month with Adwords rates). This ... conventional strategy.
As far as I can tell, you will only be able to check the SEARCH NETWORK AdWords rates for ads.
In my sector, CONTENT NETWORK ads are ending up MUCH lower priced than the search network ads as many of the big spenders opt out of the content network - then all the "low-ball" ads go to the content network. Since we all have content network ads running on our publisher pages, we may all be looking constantly at substantially reduced rates (as opposed to the search network rates) on our publisher pages. There will, of course, be a few of those big spenders who do not opt out of the content network and sometimes publishers get lucky ;).
you will only be able to check the SEARCH NETWORK AdWords rates for ads. In my sector, CONTENT NETWORK ads are ending up MUCH lower priced
This is a very interesting point. You are right in that the traffic estimator only gives a global estimate, and doesn't split it between search and content. But how do you know that content network ads are ending up much lower price because of bids?
In all the sectors covered on my site there doesn't appear to be much difference between the traffic estimator and rates I'm getting (after taking account of Google's share). Where there has been a difference it has mainly been due to my pages being smartpriced downwards (smartprice doesn't apply to search). Ie: how do you know the content rates are low in your sector because of low bids rather than a smartprice impact?
An alternative approach, that obviates the need to examine rates in Adwords, would be to compare pages that target the same keywords on your own site. The pages that have the lowest average EPC over the last month (particular if they also have high volume) are the ones to concentrate on for improving their EPC (step 2 of my methodology) or if that doesn't work by removing ads altogether and putting links to other, better performing pages (step 4).
...how do you know the content rates are low in your sector because of low bids rather than a smartprice impact?
Smart Pricing affects only PUBLISHER ads - the content network.
When I bid in AdWords, I often look to bid a "low ball" price (arbitrage). Many of these ads that were low bids will almost never show up in the search network - according to the Google reports - but will be shown hundreds or even thousands of times during the day, everyday, in the content network. This means that when MY ADS are clicked on in my fellow publisher's pages (the content network), the best those publisher's will get is a payout that is 1/2, 1/5, maybe even 1/10 or worse of the top paying ads for those keywords. This is without any Smart Pricing adjustment.
Alternately, when I adjust the low bids upward enough to compete with the "big dogs", the ads will begin to be shown in the search network as well as the content network because I have opted to be in both networks. This means that when MY ADS are clicked on in other publisher's pages (the content network), those publisher's will get a payout that is at or close to the top paying ads for those keywords.
Basically, what I am saying is that top paying ads go to the search network (where Google gets to keep all the money) and arbitrage stuff goes to the content network (where Google provides a generous split to publishers). Not exclusively, of course. I'm sure Google mixes 'n matches enough so that the content network gets a shot at some big clicks throughout the day. But overall, since around the first of this year, I believe this is Google's strategy for the ad network and SOME, maybe many publishers are going to have to learn to live as - pardon this expression - "bottom feeders" in the Google ad network.
But how do you know exactly what your worst performing pages are even if you use channels.
Adsense has 200 channels right? And your site has 2,000 pages so that won't work, there are 1800 pages you can't track.
So you use url channels.
Your site has 10 different subsections, www.yousite.com/apples/ www.yoursite.com/oranges/ etc and each section has 200 pages.
So with url channels you know that the Apples section does 3 times the business of Oranges but Oranges still earns you good money.
But within each of these sections there will be very good pages and very bad pages. But if you can only track the folder just how do you determine which pages are good and which are bad performers.
Am I missing something because I've never been able to work this problem out unless you have a site with less than 200 pages and some here have sites with 10,000+ pages.
This is without any Smart Pricing adjustment.
Everything you say fits in with my understanding apart from this last statement. It is the fact that publisher sites are subject to smartprice, and search isn't, that you are able to buy cheap ads from the content network but not from search.
I don't believe you can claim that what you are seeing is "without any Smart Pricing adjustment" because you can't tell whether the sites on which your ads are appearing have been adjusted due to smartprice or not. My contention is that it is smartprice - whether on individual pages and/or site-wide - that lowers the entry threshold that enables cheap ads to enter the site. I don't have any arbitrage ads appearing on my site though I know there are many operating in the sectors I cover.
If every single publisher followed my strategy (though in reality that wouldn't happen), I think you would find that either that your content ad displays would disappear, or that you would have to increase your bid significantly.
In fact, comprehensive use of my strategy would make it harder for those who make a profit out of arbitrage via content sites because it would significantly increase the minimum bids required for content pages. It would also increase the profits of those with original content/added value sites who receive natural traffic.
Am I missing something because I've never been able to work this problem out unless you have a site with less than 200 pages
Use the pareto rule [en.wikipedia.org] - focus on the high volume, poor performing pages. Consult your log file and set up URL channels for the 200 most frequently visited pages that have adsense on them.
I use another trick for directories where, typically, I only need a couple of channels. Eg: if I have a one or two "index pages" and several other pages with similar format/content, then I set up one channel for the directory and one for each of the index pages. I can then calculate the performance of the "other pages" simply by subtraction of the individual page stats from the directory stats.
But life would be so much easier if Google gave us automatic reporting of all URLs.
It is the fact that publisher sites are subject to smartprice, and search isn't, that you are able to buy cheap ads from the content network but not from search.
Hmmm..., blue... in an auction-style environment with a lot of bidders there will always be a top-to-bottom range of bids; and there will always be the folks who will bid on the cheap. This will happen with any system, not just Google's and not just with a Smart Pricing system.
In the Google Ad Network, in order for Google to make any money off anyone, Google must show the ad - even the low ball ad. But why would they show low-ball ads out of an inventory of high and low bids on the search network when they can cherry-pick the best for themselves, keeping all the money and avoiding the split for the best-paying ads? So Google would naturally show the best high priced ads on search and send the low-ball ads to the "discount network" - that is, lesser-paying ads show on the content network. Smart Pricing would have no role in this strategy.
I don't believe you can claim that what you are seeing is "without any Smart Pricing adjustment" because you can't tell whether the sites on which your ads are appearing have been adjusted due to smartprice or not.
What you say is true but I don't think it matters if Smart Pricing is being calculated for the pages/sites/accounts or not for this discussion. If the page in question is showing MY AD for which I BID 3 cents, then Google will not charge me more than 3 cents for any click for this search word no matter where it shows. The publisher on whose page my ad resides will only get, at maximum, the standard Google/publisher cut of the 3 cent click. If my 3 cent-bid is in a sea of inventory that has maximum bid prices up around 30 cents (as shown by the Google keyword tool), then the publisher thinks he/she got ripped by Smart Pricing. But it was simply my low bid coming in at 1/10 of the top bids.
If every single publisher followed my strategy (though in reality that wouldn't happen), I think you would find that either that your content ad displays would disappear, or that you would have to increase your bid significantly.
Agreed, not every publisher will follow your strategy and in my mind the most likely consequence for those that do is that they would soon find that less and less ads are showing on their pages. I will always have publishers showing my low-ball ads (unfortunate though it may be).
In fact, comprehensive use of my strategy would make it harder for those who make a profit out of arbitrage via content sites because it would significantly increase the minimum bids required for content pages. It would also increase the profits of those with original content/added value sites who receive natural traffic.
While I can agree and really DO sympathize with your desire to make more money per click and reduce the amount of low-ball ads, IMHO it simply isn't going to happen. The reality is that given a lot of advertisers AND a lot of publishers, Google WILL find a way to show even the lowest-bid ad and take a cut of it - just like any business is supposed to.
I don't think it matters if Smart Pricing is being calculated for the pages/sites/accounts or not for this discussion. If the page in question is showing MY AD for which I BID 3 cents... it was simply my low bid coming in at 1/10 of the top bids.
If I understand your reasoning correctly (from this and previous posts), you are saying that because you bid 1/10 of the 'going rates' in Adwords, and because you appear on content sites and not search, this shows that other advertisers are putting high bids in for search and low bids in for content.
I understand how you arrive at that conclusion, but I'm suggesting an alternative explanation that does not require a differential between search and content: that something is happening on publisher sites that opens the door for your low bids. Once there, your overall ad performance (relatively low bid, relatively high CTR) means that your low bids leapfrog the higher bids. This is due to one or a combination of factors, such as smartprice, too many adverts on the page, low average CTR on the page etc..
Whether your explanation or mine is correct, we'll probably never know. But they are two different perspective that publishers can choose from. From your point of view, the accuracy of my perspective is immaterial - the strategy I've outlined is not going to do you out of your arbitrage business, because in reality only a very small number of people, if any, will follow it. However, if my perspective is accurate then it offers publishers a way to close the back door to low bid adverts and reopen the front door to high bid adverts. It seems to have worked on my site, the only question is whether it can work for others as well.
The second step is to try to improve EPC on the worst-performing pages.high performing.... low performing
Hmmm. 21_blue, how do you know that your lowest EPC pages are the lowest converting or are you just assuming that if EPC is low conversion must be low too? So you knock out ow traffic, low EPC pages (which happen to be your highest conversion pages) and actually get your whole account smartpriced downwards?
If I understand your reasoning correctly (from this and previous posts), you are saying that because you bid 1/10 of the 'going rates' in Adwords, and because you appear on content sites and not search, this shows that other advertisers are putting high bids in for search and low bids in for content.
blue... I don't know *what* other advertisers are doing. In response to your previous questions/statements I simply stated that when I low-ball bid, why I think the overwhelming impressions of MY ADS show on the content network and why I think it may have nothing to do with Smart Pricing.
I understand how you arrive at that conclusion, but I'm suggesting an alternative explanation that does not require a differential between search and content: that something is happening on publisher sites that opens the door for your low bids. Once there, your overall ad performance (relatively low bid, relatively high CTR) means that your low bids leapfrog the higher bids. This is due to one or a combination of factors, such as smartprice, too many adverts on the page, low average CTR on the page etc..
You may be correct and in my effort to simplify things in life, I may have made this too simple. However, My bids, whether high or low, receive the same, mundane verbiage for the ads and CTR's can be quite low. It's not likely that any of my low-paying ads are going to out-perform with high click rates and oust any other. I am definitely not the classic MFA strategist.
My ads don't show up over other, better paying, well-targeted ads that can be shown, either. Rather, I think they are in a complex Google "ad soup" and show up because there is a huge Google content network and advertisers have budgets and when it's time for Google to rest the other guy's higher paying ad so that the advertiser doesn't go over budget at 4 AM or use up all his budget on 1 or 2 higher paying keywords, mine get shown.
Google is REALLY good at using all of my budgeted money for the day AND keeping my ads in play throughout the entire day. Even if I remove some of my top bidded keywords Google will show more and more of my lower paying ads in an effort to get all of my daily budget. It keeps finding publisher sites somewhere in the content network on which to run my ads.
Its this aspect of the ad network that I think most people overlook when they try to figure out *why* ads are shown - the ad budget.
21_blue wrote:But life would be so much easier if Google gave us automatic reporting of all URLs.
The increase from 100 to 200 active AdSense channels was a great improvement but, at least, if not unlimited URLs, the limit should be something like the Chitika's 1000 active channels per account, to track the performance of the most visited pages. Given the fundamental importance of testing and tracking to improve revenue, it's difficult to understand the low channel limit at AdSense, of course unless Google's computing resources are much more modest than Chitika's. ;)
are you just assuming that if EPC is low conversion must be low too?
Yes, that is the assumption that underlies this strategy. But I'm defining "low EPC" in a very particular way - by comparison with the current adwords rates for topics relating to that page.
For example, suppose I have 4 pages on two topics at my website. In Adwords, topic 1 requires a typical CPC of $1.50 to get in the top 3, topic 2 requires a typical CPC of 20c to get in the top 3. And suppose my EPC stats for the last month are:
Under my definition, pages 2 and 4 are "low EPC pages". There could be other reasons than smartprice to explain the low EPC (eg: far too many ads on the page, poor page loading sequence, etc.). But after all those factors have been eliminated (in step 2 of my methodology) if the EPC still remains at a low level then the only explanation left is that the advertiser is getting a discount from ads on that page - which must be due to Smartprice (the Adsense description of smartprice clearly suggests a page-related factor).
The low EPC may be due to low conversion, but it may also be due to the content of the page. In either case, the aim of the methodology is to remove the depressive impact of smartprice, whether that is caused by conversion or not.
my assumption is that if EPC is low (after trying all other methods of improving it) then the page is being hit by smartprice
But I'm defining "low EPC" in a very particular way - by comparison with the current adwords rates for topics relating to that page.
Also, there has been speculation, and several pages of webmaster discussion, about the account-wide effects of smartpricing. I fail to see how you are isolating this widely accepted, account-wide penalty/reward to individual pages. Or indeed on what basis you are claiming that smartpricing is rendered on a page-by-page basis. If it is not your claim that individual pages are rewarded/penalised by smartpricing then I fail to see how you can identify and remove those pages to increase your earnings.
Dangerous generalisation.
"Dangerous"?! Well, if making more money is dangerous then perhaps. This strategy is making me more money.
And whilst we can never know for sure what the relation is between EPC and smartprice (and systems thinking suggests that even Google won't fully understand it themselves) it is not simply a "generalisation", but something that has been worked out over a long period of research.
None of the methods we have for optimising income from our sites are 100% guaranteed. Saying that this strategy I'm describing is a "dangerous generalisation" is an overreaction. It is simply a strategy that you can try for yourself, and decide for yourself whether to use. And if it doesn't work for you, just go back to what you did before. Where is the "danger" in that?
The only possible danger is that it works, and then gets used widely by content publishers, and MFAers find that their income suffers as a result.
there has been speculation, and several pages of webmaster discussion, about the account-wide effects of smartpricing. I fail to see how you are isolating this widely accepted, account-wide penalty/reward to individual pages. Or indeed on what basis you are claiming that smartpricing is rendered on a page-by-page basis.
As far as I know, noone has claimed (have they?) that Smartprice contains only one factor, ie it is solely an account-wide calculation. Yes, there is an account-wide calculation, but the general view, that you refer to, is that there are many factors.
That there is also a page element is something Google have stated explicitly:
For example, let's assume you're running an AdWords ad for digital cameras. One of our content partner sites displays your ad. The first page of the site is an article about photography tips; the second page offers digital camera reviews. Because digital camera reviews is more product-specific than photography tips, a user who clicks your ad on this page is more likely to make a purchase than a user who clicks your ad on the photography tips page. In this example, Google will reduce the price of the click on the photography tips page because the estimated value is lower.
What they don't say is whether this page-related element of smartprice is based purely on conversion data or is somehow derived from content. However, it is clear from this that there is a page-related element to smartprice as well as an account-wide and/or site-wide element.
Basically, what I am saying is that top paying ads go to the search network (where Google gets to keep all the money) and arbitrage stuff goes to the content network (where Google provides a generous split to publishers).
Some very interesting and knowledgable comments, Billcale.
However, I am fairly sure that Google has been trying to keep milking cow satisfied in the first place.
In other words, there is no reason not to show top paid ads on a content network site, if the site has proven track record (conversions).
And that's where the smart pricing comes into play.
BTW, very nice thread Blue_21.
However, it is clear from this that there is a page-related element to smartprice as well as an account-wide and/or site-wide element.
21_blue wrote:3. which pages are being hit hard by smart price. You can work this out by identifying which pages have a very low average EPC by comparison with the Adwords rates
EPC = PS * CPC, where 0.75 < PS < 0.80 approximately.
Your experience is very interesting, because you are trying to identify the specific pages involved in the smart pricing effect and probably causing a secondary effect on the rest of the site and the account. Of course the discount on the CPC and EPC is the most direct measurement of smart pricing. But this method takes time to do in a page per page basis, because we don't have an "estimated conversion" column in the AdSense reports, and we have to research AdWords rates for many keywords, etc. However, it's a very remarkable method indeed.
Other different methods [webmasterworld.com], that we can call symptomatic therapies ;) , use indirect symptoms to detect and try to improve the smart pricing effect, considering that a smartpriced page usually becomes a low-performing page, with low EPC and possibly low CTR if the smartpriced page receives less quality ads (this point is controversial). Naturally there are other possible causes for a low-performing page apart from smart pricing, so those indirect methods are far from perfect.
In any case, apart from the possible role of CTR and EPC in smart pricing (controversial in the case of CTR), I think it's useful to remove ads from pages with very low CTR and EPC, and therefore eCPM (given that eCPM = CTR * EPC * 10) just to improve the user experience. This is so because low CTR ads are ads that users don't like and don't click, and low EPC ads are often ads that don't convert, and therefore ads that users find not useful after clicking.
As I've said on another thread [webmasterworld.com]:
There is a controversy about a possible beneficial smart pricing effect from removing non-performing ads, but in any case -even without that possible effect- your readers will be grateful to see only the ads that they like and therefore click on, with frequent conversions from truly interested users. And happy users are the key to more and good traffic. ;)
This is so because low CTR ads are ads that users don't like and don't click, and low EPC ads are often ads that don't convert, and therefore ads that users find not useful after clicking.
In my experience, ads on low-performing pages have perfectly decent EPCs; they just have minimal clickthrough rates due to the type of page content.
I'm thinking specifically of photo galleries on a travel-planning site. People who are looking at photo galleries may not be in a shopping frame on mind, and even if they are, they're probably viewing photo pages in rapid succession and aren't paying much attention to the ads. Hence the low CTR and eCPM. However, if those readers do see and click on relevant ads, there's no reason to assume that they're poor prospects. The fact that my photo-gallery pages deliver solid EPCs suggests that readers who click on such low-CTR pages can convert quite well.
Disclaimer: As the expression goes, your mileage may vary.
it's not only possible, but common, to have a low CTR and a high advertiser conversion rate. It's also possible to have a high CTR and a low conversion rate. Don't remove the AdSense code from your site just because it has a lower CTR - it may be one of your best converting sites.
Imho, the danger of low CTR is that it can open the door for junk ads. But if your site is not afflicted by junk ads, removing ads from low CTR/high EPC pages could simply be losing earnings.
the Adsense blog you linked to earlier provides a clear answer
However, in my case, when looking at the variations in the AdSense graphics, there seems to be that unexplained indirect, moderate relationship between higher CTR and higher EPC, but it's not completely clear, and evidently we cannot consider CTR the only or the most important factor at all.
removing ads from low CTR/high EPC pages could simply be losing earnings.
As I suggested, even in the case that there wasn't any smart pricing benefit from this, I would still remove the non-performing ads, because clearly the users are not interested in them, and users are first.
Yes, that's part of the controversy. They say "The percentage of clicks that convert for an advertiser is the most important factor"
Sorry, I'm not clear what the 'controversy' is that you are referring to, then. The statement you quote says that conversion rate is most important to advertisers - that's not controversial is it? Or were you agreeing with my point - ie: that the blog makes it clear that conversion rate has nothing to do with CTR?
The controversy was about a possible role -even indirect- of CTR in smart pricing, and quite a lot of people suggested to remove ads from low CTR pages. Instead, others like me removed ads from very low eCPM pages, since eCPM includes EPC and therefore indirectly smart pricing, and also includes CTR and therefore another approximate measure of user satisfaction about the ads (as said, eCPM = CTR * EPC * 10). Naturally, in the best case, these are very indirect methods, not very efficient.
As I said, your new method of comparing effective EPC with expected EPC (from AdWords rates for each page keywords), although a laborious procedure, seems excellent to identify directly the pages being hit by smart pricing, and therefore to take effective measures for improvement.
The controversy was about a possible role -even indirect- of CTR in smart pricing
Thanks for the clarification.
Fwiw, my data correlating CTR and EPC shows a different pattern to the variations you see in yours: average EPC is the same for virtually all CTRs, apart from a slight fall off at very low CTR (eg: under 1 or 2%). This doesn't help explain your figures, but perhaps suggests that it may not be a universal pattern.
Also, low CTR/high EPC pages contribute a relatively high proportion of my earnings, so for me it is worth sifting pages that are low EPC from those that are low CTR. But for some, as you suggest, perhaps removing low eCPM pages may be good enough because of the make-up of their site.
Thanks for your interesting and useful points - especially the reminder to take account of Google's share!
On the other hand I'm not sure if link units are affected by whether they are the first ad on the page or not. I find my link units have a lower CTR but seem to get about the same EPC as the top ad unit. This was not true for an ad unit I had at the bottom of each page. The CTR for that unit was a little lower than either the top regular ad or for the link unit.
I'd like to hear from people who use link units on how they work into the mix.
I'd like to hear from people who use link units on how they work into the mix
I don't use ad link units (ordinary ads produce much better results on my sites) so this isn't something I can comment on. Also, I don't know how the EPC-based strategy would work out with ad links, nor whether/how link and search ads are smartpriced. Perhaps someone else can comment on their experience with adlinks?
Adlinks work well for me, especially where the screen real-estate that I can spare above the fold is limited.
I occasionally run adlinks below the fold, where the page design doesn't really allow anything else, and they bring some income.
For me Adlinks do at least as well as normal ad units I think, and better considering the pixels they need to deliver the same $s.
Rgds
Damon
