Forum Moderators: martinibuster
Smart pricing doesn't work at all as described or intended. If it did the results wouldn't be so seemingly random or fluctuate so wildly.
I haven't seen wild or seemingly random swings that I would attribute to smart pricing. On a daily basis, I do see some fairly large swings, but I can usually connec them to changes in traffic or in advertisers. And those swings have died down as my overall AdSense impressions have grown.
So I'm wondering about this. If the smart pricing factor is updated weekly, as has been suggested here, then its impact should only be noticeable from week to week, and not day to day.
What have other people seen? What kind of changes in earnings or EPC has smart pricing caused, and how do you know it was smart pricing, and not something else?
And what kind of site(s) do you run? That may have an influence--I run a small, fairly focused, content site. Perhaps that kind of site is less vulnerable to fluctuations...
I don't want to sound snarky here, but a) Do you know what smart pricing is in its detail? b) Do you know how it functions specifically? c) Do you know how and when it is applied?
The answer to all three questions is of course not. The answers are also irrelevent given the discussion at hand. I am well aware of the complexity involved in the production of Google's algorithms, and true, every engineer likely doesn't know how every part of them work. But the team as a whole had better know how the end result will affect any given site or the results will in fact be random for all practical purposes. We outside of Google don't need to know the ins and outs of their algos to know when they're working or not.
The premise behind smart pricing is simple even if its implementation is not: What is the likelihood that a click from a given page will result in a conversion? It's quite obvious from the evidence at hand that the algo has no clue.
The premise behind smart pricing is simple even if its implementation is not: What is the likelihood that a click from a given page will result in a conversion? It's quite obvious from the evidence at hand that the algo has no clue.
That may be YOUR idea of the premise, but is it googles? How do you know? It's certainly a huge simplification, which would suggest to me that it's quite off the mark. Actually the premise I would suggest doesn't have to do with individual pages, but pages over the aggregate. That is, whether it is accurate or not for your widget page is not a practical concern for google. What IS, is that over the network, the algo can do what it is supposed to.
But in fact, its actual accuracy is largely irrelevant in a business sense (or at least it could be), since whatever they are doing is probably oriented towards building ADVERTISER confidence. PERCEPTION is the operative word, and it's probably only a secondary concern that it's fair, or that it pleases webmasters. I'm speculating of course. We don't know what's in the black box, or why the black box was really created.
Thus, it can DO exactly what it is supposed to in the aggregate, while not appearing to do it well for your widgets page. It's like actuarial tables. What may look broken for you may mean it is working well for google.
Of course, to make your comment sensical, you'd have to somehow demonstrate that across the entire google system "smartpricing" does not predict conversion. Can you do that? I doubt you have the aggregated data to do so. If you had the predictive mathematical model that smartpricing uses to make predictions (including all the variables and their interactions) and the data, you'd have a shot at judging smartpricing broken.
That may be YOUR idea of the premise, but is it googles? How do you know?
Because Google says so:
So if our data shows that a click from a content page is less likely to turn into actionable business results -- such as online sales, registrations, phone calls, or newsletter signups -- we reduce the price you pay for that click.
Thus, it can DO exactly what it is supposed to in the aggregate, while not appearing to do it well for your widgets page. It's like actuarial tables. What may look broken for you may mean it is working well for google.
This is in direct opposition to Google's explanation that smart-pricing is in fact page-based. (Please re-read their paragraph above if this isn't clear by now...)
If you had the predictive mathematical model that smartpricing uses to make predictions (including all the variables and their interactions) and the data, you'd have a shot at judging smartpricing broken.
No, I only have to prove that smart pricing does not in fact accurately predict whether a click from a page is likely to result in a conversion. And since Google obviously cannot make a concrete decision in that regard themselves I have done so IMO.
This is in direct opposition to Google's explanation that smart-pricing is in fact page-based. (Please re-read their paragraph above if this isn't clear by now...)
Well, of course it's APPLIED on a page level. But whether it is broken or not is determined by the accuracy over the aggregate.
The best way to explain it is actuarial life insurance tables used to calculate rates, etc. The variables they consider result IN THE AGGREGATE, in the insurance company paying out their benefits in a way so they stay in business/make a profit.
Their "algorithm", if you like, would be "broken" if it resulted in the company having to pay out more money than it could afford (or reduce profit below a certain level...whatever the criterion). OVERALL.
Actuarial tables, however, will not guarantee that the company will make a profit from your Uncle Eyegor. He may, for example, be 30 years old, non-smoker, no risk factors, etc, and may sign up, and pop off three weeks after buying the insurance. The company takes a HIT. Are the actuarial tables "broken? No, of course not. The insurance company, overall, doesn't really care whether your Unc. beat the tables.
The insurance policy applies to the individual (UNC Eyegor), but whether the predictive algorithms work is ONLY determined in the aggregate. Whether it is broken can only be determined by looking at the aggregate, because, it was DESIGNED to work IN the aggregate. It was never designed to fit/work with your UNC.
It's probably certainly the same with smart pricing or, for that matter, any google algos. Whether it's CPC, search positions, spam issues, etc, the ONLY way you can deal with a wide range of situations, is algorithmically, and evaluate IN THE AGGREGATE.
It's probably certainly the same with smart pricing or, for that matter, any google algos. Whether it's CPC, search positions, spam issues, etc, the ONLY way you can deal with a wide range of situations, is algorithmically, and evaluate IN THE AGGREGATE.
And since these wild fluctuations take place on multiple sites within one account it is reflected in the aggregate. The problem is affecting the entirety of both sites, not just individual pages. But the errors have to occur on a page by page level before they can be "aggregated" - otherwise Google's description of how smart pricing is supposed to work cannot apply.
What have other people seen? What kind of changes in earnings or EPC has smart pricing caused, and how do you know it was smart pricing, and not something else?
Sorry you're regretting the thread. The straight up answer, honest to god no bs, is that nobody here can do anything but speculate and pretend their conclusions are accurate. It's that simple.
And, there's the "failure to communicate" that occurs, when the word "aggregate" as I used it, is interpreted to mean over two frickin sites!
Anyway, I'm done. Thanks all, it's a stimulating topic. Fun. Pretty much of no use at all. Like eating the red smarties first. Or, kind of like complaining about the weather, although at least with the weather you can bundle up. If anybody "got" what I was saying, great.
But what I was hoping to hear about was other people's experiences--concrete information, not theoretical argument.
I like your "actuarial tables" analogy, and I think it fits nicely with Google's description of how smart pricing works.
I also agree that smart pricing takes the blame for a lot of things that may not even be related to smart pricing. One thing I've speculated on in the past is Google's AdSense compensation formula, which could very easily be designed with greater rewards for certain types of content than for others, a sliding percentage scale, etc. Also, it makes sense that Google would try to spread any given ad over as many sites as possible (or at least a good number of sites) instead of letting a handful of sites monopolize clicks for that ad. By doing so, Google would be spreading the wealth around for publishers, but--far more important--it would be minimizing risk for advertisers who might be worried about click fraud or poor conversions from high-traffic but low-quality sites.
But what I was hoping to hear about was other people's experiences--concrete information, not theoretical argument.
Hunderdown, I understand what you are asking for. What I'm saying is that the only source for concrete information that would be useful is google, and they have a vested interest in nobody having that information.
This is exceedingly practical to understand. Let me give you an example from google search. Then it's bed, and I'm finished on the thread.
Imagine that you want to know how to get better search placement in google. Someone on webmasterforum says: "Heck, build content", so you go to an articles repository and grab their legally reproducable content and add it every day to your site. But guess what? Because you didn't know how the black box of google worked, you also didn't know that the content you added was seen as duplicate from google's point of view, and hence, all your effort was largely wasted. (this isn't quite so strictly true about dupe content - I'm simplifying and removing the "it depends" variables on this). Understand also that we didn't always "know" about dupe content, and we actually lack a lot of info about how it is determined, etc, but at least we know a little, sorta.
The very PRACTICAL point I'm making is that we need to separate our opinions and speculations from how things "really" work, particularly when we can't possibly know what's going on in the black box. Knowing what we don't know, and knowing what we we probably will NEVER know is important in a business sense, in a very practical way.
Also, knowing why a particular argument is false IS practically important. And sometimes, you have to understand the bigger picture to understand why a particular conclusion is faulty (in this case you really need to understand the reason WHY we can't ever know about smart pricing is that a) we aren't meant to and b) we don't have the data across the whole network. What happens on my site won't tell you anything. Or if you tell me about yours.
Finally, I'd figure that about 95% of the discussions about google -- search or adsense are sheer speculation but often presented with a veneer of fact that can be easily pierced by a little understanding of bigger picture issues.
Understanding algos and aggregate issues is necessary to separate the huge majority of simple speculations from the gold that might make you some more money.
I could give you information about my sites re: your specific questions. The thing is that I have NO data that will REALLY answer any of your questions that make inferences about how google works, or whether something is broken or not. All I can give you is info and tentative conculusions about MY sites, which is virtually worthless in terms of a) YOUR sites, and b) how smart pricing really works.
There are tons of mysteries associated with Google, and I'm sure that's deliberate so that the system is hard to "game". We know SO LITTLE that very few real conclusions are possible, and most of the data we have access to, while useful to the site owner whose data it is in a limited way, has almost no "probitive" value for anyone else on the planet. And their conclusions based on their sites, no matter how large, are even less valuable, since not only are they based on inadequate data, but the logic used is often borne of ignorance (simply not knowing), and faulty.
Like most, it kinda drives me nuts. I want to know, too. I'd love to know why things fluctuate (or don't) on my sites). I'd love to know why, no matter what happens (higher traffic, more impressions, less impressions, whatever) overall, I almost always end up around the same revenue. Or why it is almost inevitable that on any given day, half my sites will improve performance over yesterday (CPC, CPM), while the other half will drop, again leaving me exactly where I was yesterday. And dozens of other questions.
But I don't know, I won't know, and no matter what others say about their sites, I still won't know. It's very practical to know where to spend one's time. If you want an entertaining convo, these questions are great. But if you run a business, it's very practical to understand that your questions simply can't logically be answered from here. And never will be.
Simply, you can't get there from here.
Trying to figure out what you can never figure out about google can drive you nuts, and wastes a lot of time. But it's cheaper than going to a movie.
I am not concerning the actual algorthm and I would say don't spend effect on the AS optimization as it keeps rolling in minutes, that is the way Google should go or there business will be crashed because of being cracked.
P.S. In my case, there is another major effect in Feburary, the averaged EPC and income drop about 50%.
The principal reason behing smart pricing is to make advertisers think they are getting a discount, and to increase profit for Googles investors.
I have no problem as a publisher in smart pricing discounting clicks. There are many scenarios I can see where a click could legitimately be said to not convert. The problem I have with smart pricing is that it doesn't feed the discount back to the advertisers. If it did then I'd be OK with the stupid algo.
That means my site is considering to be less "valuable" under AS and too that means AS is coming less important than other tools in monetizing my total income ........ :)
The smartpricing effect to me as a publisher only has one meaning.
That means my site is considering to be less "valuable" under AS and too that means AS is coming less important than other tools in monetizing my total income ........ :)
Excellent post!
Smart pricing or not...at the end of the day it is all about the $$, also for us publishers.
Continued Bad Algoritmic (like) earnings = search ad code & replace. It is that simple.
Smart pricing or not...at the end of the day it is all about the $$, also for us publishers.
Continued Bad Algoritmic (like) earnings = search ad code & replace. It is that simple.
There's a tendency for publishers to forget several all-important realities:
1) Advertsers (not publishers) are Google's customers.
2) Google's first responsibility is to deliver full value to those advertisers.
3) The publishers who quit AdSense because of poor earnings are likely to be the publishers who are least valuable to Google. From Google's point of view, it may make perfect sense to let low-performing publishers switch to competing ad networks.
"Sorry your site does not give value to our advertisers, you are being thrown out of adsense"
As others have said, advertisers are the customers. You must be value for money for them.
if that was the case, the smart business decision would be to simply cut 'em off completely, which hasn't happened, afaik.
for all you know, google has been taking more of the pie for itself, because it's stock prices took such a beating recently... many publishers have had their epc cut, as we have seen in numerous posts these last few weeks.
so the smart publishers move on to better ad networks... if there is such a thing these days, lol.
That is not the point.
All REASONABLE publishers welcome any measure from Google to filter truck sites. However, the reality in front of my eyes is the Smartpricing is not implemented effectively.
More important, publishers are looking for tools and statistics to help for their individual optimization for "perfect matching" the suitable Advertisers.
Is that because publishers are NOT customers and they DON'T PAY so publishers do not have the choices?
That is the reality. From the publishers points of view, the smartpricing makes AS becomes a less competitive monetizing tool, and Google did not provide additional tools for the publishers, the net result is obvious.
Google probably has overweight the smartpricing, they lose the point, they would better improve their optimization tools for publishers instead of smartpriced the publishers.
if that was the case, the smart business decision would be to simply cut 'em off completely, which hasn't happened, afaik.
There's no need to cut publishers are as long as they're yielding a profit for Google. But there's no point in Google's taking a loss on them (and driving away advertisers) by paying more than their traffic is worth.
for all you know, google has been taking more of the pie for itself, because it's stock prices took such a beating recently... many publishers have had their epc cut, as we have seen in numerous posts these last few weeks.
i've heard that rumour, but i've never seen specific payout numbers in their quarterly earnings report... got a link? and fyi, i've never stated that google is "cutting the payout".
as for the latest quarterly earnings report: "Google shares plunge as earnings disappoint" ft dot com, 1/31/06
Scott W. Devitt predicted that google's stock was going to take a dive, as reported at the washingtonpost dot com on 2/6/06:
"...Since Google reported its earnings, analysts have finally started taking a harder look at the company's numbers. Google blamed the disappointing profit on income taxes, saying it wound up paying a higher tax rate than expected.
Devitt rejected that explanation immediately, saying that even without paying more taxes, Google would not have hit its profit target. By Friday, the Wall Street Journal was talking up the same theory, calculating that taxes accounted for only half the disappointment, and quoting Devitt.
Wall Street also is likely to start listening to some of Devitt's other concerns.
He thinks Google advertising is infected with "click fraud," a term that covers a variety of ways that advertisers, their competitors and others can game the system and manipulate the number of "hits" that online advertising attracts. He thinks Google's advertising rates are headed for a fall because advertisers aren't getting the results they want."
after months of steady earnings that were on a constant upward trend, my epc went in the toilet in the middle of january... and it still hasn't returned to where it was... google knew that they were going to take a big hit.
it's really funny when people call that "smart pricing" :-)
after months of steady earnings that were on a constant upward trend, my epc went in the toilet in the middle of january... and it still hasn't returned to where it was... google knew that they were going to take a big hit.
I guess if all you look at is YOUR site, and ignore the fact that some people have experienced an INCREASE from adsense, all kinds of conclusions are possible (and potentially amusing).
My earnings are down too as I said before, but I don't feel the need to attribute this to SmartPiercing(TM), especially as non-AdSense CPM revenue dropped in sync.
Remembering the difference between "good" and "good for me" is important if you intend to stay rational and fix whatever is ailing you, or as Dogbert would say: "It's not always about you."
Actually, though it's waaaay too early to say yet, I seem to have been able to tweak my AS eCPM upwards again though the other (CPM) networks' eCPMs are staying put for now.
Rgds
Damon
PS. hunderdown, thanks for starting this thread...
i've heard that rumour, but i've never seen specific payout numbers in their quarterly earnings report
Just go to any Google quarterly earnings report and divide the AdSense partners' revenue share (in dollars) by total AdSense revenues. That will give you the average payout percentage.
He thinks Google advertising is infected with "click fraud," a term that covers a variety of ways that advertisers, their competitors and others can game the system and manipulate the number of "hits" that online advertising attracts. He thinks Google's advertising rates are headed for a fall because advertisers aren't getting the results they want."
Actually, most advertisers are getting the results they want. In another thread (possibly on the Supporters Forum?) there was a link to an article that compared the average cost per lead for various media. PPC was a bargain compared to print ads, direct mail, etc.
Sure, click fraud is a problem, but as long as it's kept to levels that don't cause poor ROI for advertisers, it isn't a significant factor in whether Google and other PPC networks attract or don't attract ads. What's more, pay-per-click is only one pricing model, and I expect we'll see a lot more CPM in the years ahead.
after months of steady earnings that were on a constant upward trend, my epc went in the toilet in the middle of january... and it still hasn't returned to where it was... google knew that they were going to take a big hit. it's really funny when people call that "smart pricing" :-)
Why do you assume that "smart pricing" is responsible for your woes? And even if it were, why would you assume that a drop in your EPC is an indication that smart pricing doesn't work? The goal of smart pricing is to ensure value for advertisers (thereby attracting and retaining advertisers), not to ensure that you get rich.
Smart pricing and wild speculations
And why is it that we don't seem to see many people giving Smart Pricing the credit when their earnings go up?
That's why I was asking for people's actual experiences, not their theories.
Why let actual facts get in the way of a good bout of paranoia?
And of course it is vital to extrapolate at least 3n+7 theories from n data points, and n should be 0 or 1 for the most fun! B^>
Rgds
Damon
there is no such thing as "smart pricing", lol... that why i find it so funny to see people like you constantly chatting up the concept.
and no, the google quarterly report does not tell you what the actual payout data is.
what might be helpful here is some kind of a poll, just simple choices with no comments... either your epc went down in january, or it went up.
