Forum Moderators: martinibuster
This month Google added a revised method of calculating the price of ads based on the "effectiveness" of the ad.
Here's an excerpt:
-------
How smart pricing works
We are constantly analyzing data across our network, and if our data shows that a click is less likely to turn into business results (e.g. online sale, registration, phone call, newsletter sign-up), we may reduce the price you pay for that click. You may notice a reduction in the cost of clicks from content sites.
We take into account many factors such as what keywords or concepts triggered the ad, as well as the type of site on which the ad was served. For example, a click on an ad for digital cameras on a web page about photography tips may be worth less than a click on the same ad appearing next to a review of digital cameras.
Google saves you time and hassle by estimating the value of clicks and adjusting prices on an ongoing basis. With improved smart pricing, you should automatically get greater value for clicks from ad impressions across our network, all with no change in how you bid.
------------
So it appears that our EPC will be way down from before unless we deliver more EFFECTIVE clicks. Now what I do not understand is how that is measured. I can understand how they measure click-throughs (we get as much as 3% on some days, but usually 1.8%), but how are they deciding what the visitor does once they get to a site? They do not have a way to track a purchase without a tracking code for each specific vendor. We use AdWords and we know of no such tracking code. Seems kinda like guessing and a lot of assumptions based on some marketing guru's idea of what can deliver more effective business leads and not one ounce of imperical measurment.
You may notice a reduction in the cost of clicks from content sites.And now it's official. Content is no longer king. In fact, content is now the mighty dollar's bit@#.
Google saves you time and hassle by estimating the value of clicks and adjusting prices on an ongoing basis
This is one automation too far. How can an automated system judge the value of a click? Even a human can't do that. The only way the value of the click can be gauged with any level of accuracy and fairness is by Google comprehensively tracking all referrals at the merchant's end and having access to all merchants' confidential information like ROI.
[edited by: Macro at 3:27 pm (utc) on April 11, 2004]
Perhaps they have some marketing data that shows a visitor is "more likely" to make a purchase if the click is from a product review than if the click is from a How To page.
Their "photography tips" vs. "camera review" comment was just a hypothetical example, and it's possible that it wasn't even grounded in reality. For all we know, some copywriter might have made it up to illustrate the concept of variable pricing.
I don't know how such an approach would work in a category like travel, where the majority of readers are looking for ways to spend their money. Even a reader of a "tips" page on a travel site is likely to be a prospective buyer. Take an article on how to book vacation apartments, or an article about European rail passes: neither is a review, but readers of the first article are probably looking for accommodations, and readers of the second piece are strong prospects for rail travel.
Maybe I'm naive, but I can't see how an algorithm would be able to tell whether those articles were likely to generate "business actions" unless conversion rates for the individual articles were being tracked. In the case of the railpass article, for example, would Google be able to tell the difference between an article for people who are planning travel and a personal account of traveling around Europe with a rail pass?
I'm inclined to think that, except in cases where an advertiser's conversion data is being tracked, Google is basing its assumptions on one or all of the following:
1) Content-site conversion data for the type of product or service that's being advertised.
2) Conversion data for the domain.
3) Conversion data for the general type of page or site (e.g., editorial, e-commerce, directory, parked domain, gmail, etc.) without getting into details like whether it's a "tips" or "review" page, which can't be determined reliably with an algorithm.
Google is basing its assumptions on one or all of the following
EFV, knowing Google, they'll use a lot of factors in calculating value of clicks. If they are automating the calculation of click value then it they would need to have some mathematical basis for that calculation. The type of page/page category would be a vital consideration (as in editorial/directory etc.) along with the type of site... and other factors, maybe even how often it is updated!
Conversion data doesn't necessarily come into it - Google don't/doesn't have this information. But they will set their algo to "judge" the category of the site/page based on factors like how many sites link to them with the anchor text "Review of ...", whether the site sells any products, whether it's closely linked to academic sites.... and tons more.
The scary part is that to increase the accuracy of click valuation they would need to include in the calculation factors that they say they don't consider at present. Surely these are all factors that - if applied - may increase the accuracy of click valuation:
- the number of visitors who are making repeated trips to that page (surely that must count!)
- the version of browser they are using
- the IP they are accessing it from (AOL dial up subscribers are hardly likely to be buying books on Unix :-))
- the other pages the visitor saw prior to landing on this one (info courtesy toolbar)
- where the page ranks in SERPS
...tons more. Many of them will involve using "confidential" data they have including data on how people travel through your site, who your referrers tend to be, how long people spend on each page etc.
Either - in the interest of getting a click calculation fairly accurate - they are going to use all the data weaponry at their disposal, including "confidential" data from a variety of sources like Toolbar, Adsense stats and SERPS click tracking .... or they are going to resign themselves to their automated system of click calculation being inaccurate and unfair.
Maybe we should at this stage refrain from calling this "click valuation" and call it "click value guessing" or maybe Algorithm for the Arbitary Allocation of Click Value (AAACV). Perhaps others have a better suggestion/Acronym? ;)
That is insanity as there is no way to judge that in many (if not most) cases involving certain products and services it's extremely unlikely for a direct online order to be placed no matter how great the offer, adsense or website is.
I believe on average marketing research indicates about 8 contacts are required before a buyer decides to buy, especially so with high-priced, niche products or products which are costly or risky to buy. For example far fetched to order online a $3000 commodities trading system software program right away by clicking the adsense ad, without first calling, emailing and looking at the website many many times!
How can G be so blind on the fact there is no way to realistically judge or even roughly estimate the click values with many diverse products/services?
Yeah, pages with toddler toy reviews are _much_ less likely to convert their traffic with an advertisement for anti-aging books. Is that the fault of the publisher or the ad targeting?
I can see using some conversion information to help target ads better, but to use it to judge the value of the page for advertisers overall is to ignore the largest factor in the conversion equation, Google's targeting.
ie. could it be as simple as this...
"If you are an Adwords advertiser and you sell blue striped widgets, you'll pay on a scale like this:
top ppc: someone searches for blue striped widgets and clicks on your ad. because you've used the keyword 'blue striped widgets'.
level 2: someone looks at a site about blue striped widgets and clicks on your ad.
level 3: someone searches for striped widgets....
level 4: someone visits site about striped widgets...
...and at the bottom...
someone visits a site about dotted widgets, that mentions striped widgets a few times."
Would this sort of categorisation need a sophisticated algorithm?
It's some communistic formula.. why not let the market dictate the price on a per content site basis?
As an advertiser... I think I would be extremely worried about Google charging me on the basis of a secret, undisclosed algorithm.
OK, so it's presented as a variable "discount", but even so I'd think advertisers would want more details as to how their money's being spent.
First of all, Google is not "giving" publishers anything. Remove the ads and see how fast the money stops flowing in. You are providing a service (ad space), and are being compensated. The money may or may not be better than what you could obtain elsewhere, but to call it free money or a gift diminishes what we as publishers are providing: valuable ad space in niche markets.
Second, it is my opinion that Google start providing some concrete numbers for publishers. The program has matured enough to do so, and with this recent change leaving many publishers with ill feelings towards Google would do much to restore confidence in their program.
Second, it is my opinion that Google start providing some concrete numbers for publishers.
I get concrete numbers in my reports every day. They've been crumbling a bit lately, but they're concrete. :-)
I've never heard of a middle-man who has the power to adjust his cut on the fly without telling either side how much exactly that cut is.
Google isn't a "middleman" in the way that traditional ad networks are. It's more of a value-added packager that buys a product (ad space) from us, marks it up, and resells it to advertisers.
Traditional ad networks say, "Let us handle your ad sales. We'll split the proceeds with you 50/50 [or 60/40, or whatever]."
Google says, "Unleash the full revenue potential of your Web site. AdSense delivers text-based Google AdWords ads that are relevant to what your readers see on your pages — and Google pays you."
If a traditional ad network were to hide its revenue split or change its payout formula, you'd have grounds for complaint because the and network would be violating the terms of the deal. But with AdSense, the deal isn't about a revenue percentage; it's about the size of your check at the end of the month. And you can walk away from the deal whenever that check isn't big enough.
Traditional ad networks could also keep the payout formula secret and change it whenever they wish, if the contract were written that way. And frankly, if they paid 10X what everyone else paid, there would be a line around the block to sign up.
P.S. I've virtually never run banner ads on any of my sites, but I believe that they DO keep certain things secret and they DO make changes to the payout formula at will - even retroactively.
We know Google staff visit sites from time to time, and every publisher accepted has been reviewed by a human.
Google very well may have an ongoing human review of sites, and these humans can "adjust the algo" or whatever you want to call it on a case by case basis.
All payout amounts are black boxed, so anything is possible... And the legal agreement not to discuss EPC values protects them from webmasters comparing raw data to find out what's going on.
In most businesses real estate, sports agents, stock brokers the percentages are right there for both parties to see...
This is true in all types of partnerships (and relationships).
The fact is, Google increases the distrust of its partners by not telling them what is going on. This really is true almost by definition.
Unfortunately, there is absolutely nothing we can do about it except by making sure we figure this into all of the decisions we make.
Well, there is nothing we can do about in on WebmasterWorld at least as a call to action is really against TOS.
Unfortunately, there is absolutely nothing we can do about it except by making sure we figure this into all of the decisions we make.
Precisely. It all boils down to whether the revenue from AdSense justifies running an "Ads by Google" box on your pages. It's a simple matter of dollars and cents.
I am encoaraged that GG has indicated that a discount (less CPC) will be issued for ads on content sites. This mirrors the EPC reduction publishers have witnessed.
I wonder, how many advertisers have not checked the "content site" box? is this prevalent? Is this a response to something GG knows and publishers and advertisers do not? Are advertisers more likely to advertise on content sites if they pay less?
In the real world, transparency doesn't give the opportunities for manipulation that it does online. That's a huge issue for the long term success of Adsense, because its actually visitors that will determine whether revenues flow, not Google.
I wonder, how many advertisers have not checked the "content site" box? is this prevalent? Is this a response to something GG knows and publishers and advertisers do not?
Google keeps expanding its definition of "content ads" with things like the DomainPark program and gmail. This lowers the quality of the content-ad pool, so it's inevitable that more advertisers will opt out as conversion rates and quality of leads decline. One solution would have been to give advertisers more control over where their ads appeared; Google chose a scheme of variable, automatically-applied discounts instead.
Are advertisers more likely to advertise on content sites if they pay less?
Good question. If advertisers are as suspicious of Google's discounts as publishers are of Google's payout scheme, the answer may be "no" in many cases. IMHO, advertiser skepticism is a bigger problem than publisher skepticism, because advertisers are shelling out their hard-earned cash for ads. Google is betting that advertisers will trust an algorithm to give them full value for their money. That may be a questionable assumption, since worthless traffic (e.g., traffic from DomainPark or gmail) doesn't suddenly acquire value when it's purchased at a discount.
Precisely. It all boils down to whether the revenue from AdSense justifies running an "Ads by Google" box on your pages. It's a simple matter of dollars and cents.
Let's also not forget the branding effect Google enjoys by having 1000s of sites of all kinds, all over the world, deliver its ads along with free advertising of Google itself via the "Ads by Google" box on top.
Let's not forget that (at least in Europe, where I live) Google has dominated (ON MERIT ofcourse), because of the TREMENDOUS support from the Webmaster community.
Let's not forget the influence of "computer geeks" over 1000s of "ordinary users". A recommendation by a knowledgeable non-affiliated person in a mailing-list or newsletter is much more effective than $s in paid ads.
It's obvious Google is intentionally being a bit lax on the criteria for a site to join AdSense, because they KNOW that having 1000s such small sites as AdSense partners will be a powerful advantage in the future, if/when the SE competition gets really fierce (e.g. Yahoo Slurp is already deep-crawling constantly like mad the last few weeks and I see new content in the SERPs very quickly, and Micro$oft has complete control over the end-user desktop)
I don't think it would be smart for Google to jeopardise this relationship by making Webmasters feel that they are being treated unfairly.
I know it's difficult for some to accept a much lower CPC income vs that of a few months ago, but we must also try to stay objective. I don't use Adwords yet (so I don't know what tools are used), but I'd like to see Google work really hard on calc'ing conversions-ROI as accurately as possible to award fair CPC to each AdSense site.
Having said all that, I'm "testing" Adsense for the last 3-4 days, and it seems as if CPC is off by an order of magnitude (ie a zero is missing ;-) vs what I had imagined (having some idea of the ad sector) and vs results from un-targeted non-Adsense ads. But I'll wait for a few weeks for the algo to settle a bit, before drawing conslusions.
Afterall, participation in the Adsense program is so flexible, that a webmaster is free to decide if it fits him.
Having said all that, I'm "testing" Adsense for the last 3-4 days, and it seems as if CPC is off by an order of magnitude (ie a zero is missing ;-) vs what I had imagined (having some idea of the ad sector) and vs results from un-targeted non-Adsense ads. But I'll wait for a few weeks for the algo to settle a bit, before drawing conslusions.
I know a couple of publishers with entertainment-related sites who are reporting daily income of "pennies" from AdSense. I don't know what their traffic is, but I can understand why anyone who's earning pennies a day from AdSense would be tempted to pull the code.
I don't use Adwords yet (so I don't know what tools are used), but I'd like to see Google work really hard on calc'ing conversions-ROI as accurately as possible to award fair CPC to each AdSense site.
Following up my own comments, to be fair, I just finished reading G's AdWords FAQs on conversion tracking info.
It seems to me that the AW tools to track conversions are adequate (keeping in mind the Keep-It-Simple-Stupid principle, no tech skills req) with obvious limitation, e.g. if the lead's followup is to occur by phone and not online etc. For this, I assume that advertisers actually USE the conversion tools offered by AW.
If advertisers don't/can't use the conversion tools because sales happen on the phone, what can G do? It can't directly monitor "efficiency" of a click from site A vs site B to determine "fair CPC", can it?
In the case that G conversion tracking tools are not used, G can only use supply-demand to derive CPC for a particular industry AS A WHOLE (but not value PER AdSense SITE), and has to use other factors, much less objective and reliable, to determine "quality" of a click from site A vs site B.
So I guess an AdSense publisher has to wait and see where CPC goes after a statistically significant # of clicks. And whether there are RLEVANT, quality and decently paying ads for his content.
If results are not acceptable, one can always discontinue AdSense, not much harm done (unless one gets many embarassing mis-targeted ads about gay weddings ;-)
I know a couple of publishers with entertainment-related sites who are reporting daily income of "pennies" from AdSense. I don't know what their traffic is, but I can understand why anyone who's earning pennies a day from AdSense would be tempted to pull the code.
OK I have tested AdSense for the past 10 days. Certainly a very short time, EPC has fluctuated alot, and it was after the 1-Apr changes, but now I have some idea what to expect.
The targeting of Adsense ads overall was not bad, often very good.
The CTR was improved vs untargeted ads, BUT the revenue per visitor (ie $ per impression as measured by G, NOT revenue per click) has been about 1/10 of the untargeted ads.
I've done some preliminary research of Google's own SERPs and in the sponsored listings I see the same ads that appear on my site. Just 1-3 sponsor ads in G SERPs at a time, whereas in other sectors (e.g. travel) G gets all 8 slots of sponsored links filled in their SERPs.
So, my best guess is that there are few advertisers in certain sectors, so there's little demand, and EPC is ultra low.
Advertisers get a great bargain, free exposure and clicks at rock-bottom cost.
