Forum Moderators: martinibuster
My CTR this week is higher than it's ever been before yet earnings (CPC/eCPM) dropped a bunch this week and the same CTR (a record for my site) last week would've been $100-$200 more than this week.
So much for the UPS club...
Not that I'm completely complaining, but it just seems very odd that nearly a thousand more clicks would make a lot less money all of a sudden.
I think I'll get the flash light, the can of crisco and sit in the corner and brace for impact.
Between all the sites, the poll received over 400,000 non-repeat votes - not a huge sample size but large enough to get more than a rough idea. Slightly less than 9% of the respondents use conversion tracking.I believe that the vast majority of AdWords advertisers are mom and pop operations who barely know how to run a website, let alone set up conversion tracking, and my poll results bear that out.
Google has never said that "smart pricing" discounts are based on conversion-tracking data for specific sites. Such data, where available, may come into play, but when smart pricing was introduced, Google talked about the likelihood of conversion from certain types of content, using the example of a camera-review page (which would be expected to convert well) and a page of photo tips (which would be expected to convert poorly). So it really doesn't matter if 9% or 99% of advertisers use Google's conversion tracking. What does matter is (a) whether Google believes it has enough data to make educated guesses about conversion rates for various types of content or sites, and (b) whether smart pricing is attractive to advertisers (as it appears to be, if we're to judge from the ad revenues in Google's quarterly earnings reports).
Also, smart pricing (or Google's payout to publishers, for that matter) doesn't have to be based only on conversion rates. If Google wants to, it can take other factors into account, such as what it thinks is good for the AdSense network or for the Web in general. To use a hypothetical example, if Google were to decide that scraper sites were undesirable, it could increase "smart pricing" discounts for clicks from scraper pages and/or pay out a smaller percentage of scraper-page revenues to publishers.
Supply and demand obviously play a role in the daily ups and down that we see (AdWords/AdSense is an auction-based market), but that doesn't exclude the possibility that smart pricing and/or the compensation formula influence publisher earnings.
Another consideration; earlier this year, we were just focusing on one part of a specific industry, and there were two very distinct time periods where earnings would drop around 20%. With a little research we were able to more or less determine where our demographic was going, and develop a new set of sites that appealed to their primary interest during those specific months.
I'm going to assume it is not from the same source as your base
The recent increase in traffic is coming mostly from Google so it's only as good as what Google sends me ;)
as long as you are looking at a rise in earnings, mark it up as a win.
In the grand scheme of things I can't consider higher earnings a losing thing, it's just frustrating to see my ad inventory selling for less as the clicks per day increase.
<grumples incoherent profanity>
[edited by: incrediBILL at 11:49 pm (utc) on Sep. 28, 2005]
Yes, I would. The laws of statistics predict that such things will happen, and we have periods in our stats that exhibit the characteristics you describe (and other apparent 'anomalies').
I agree with EFV that it is the long term that matters. I also think that, to understand the market, one needs some understanding of statistics and some data against which to benchmark sites' performance. Bottom-line figures result from a complex interplay of many factors, some of which you can control (eg: ad placement), some you can only influence (eg: volume of ad supply) and some are constraints that you have to live with (eg: the Adsense algorithms).
Did anyone mention YPN? I've never used AW, so correct me if I'm on the wrong track here, but it occurs to me that perhaps in an effort to keep AW users from leaving the plantation, Google just worked in a bit of a discount for all.
If that were true, wouldn't we be hearing shouts of glee on the AdWords forum? :-)
...Google talked about the likelihood of conversion from certain types of content, using the example of a camera-review page (which would be expected to convert well) and a page of photo tips (which would be expected to convert poorly). So it really doesn't matter if 9% or 99% of advertisers use Google's conversion tracking. What does matter is (a) whether Google believes it has enough data to make educated guesses about conversion rates for various types of content or sites, and (b) whether smart pricing is attractive to advertisers (as it appears to be, if we're to judge from the ad revenues in Google's quarterly earnings reports).
But it does matter whether 9% or 99% of advertisers use conversion tracking. The variations between the quality of sites is simply too great to extrapolate data from a statistically insignificant sample size onto other "similar" sites (or pages). Even pages with the exact same "types of content" can and do vary enormously in quality.
The "liklihood" that ads on one camera review page will convert has absolutely no relationship at all to the liklihood that ads on another camera review page will convert.
I'm not disagreeing with you when you explain Google's thinking about smart pricing. It's pretty clear what their reasoning is. The problem is their reasoning is obviously flawed. IMO Google simply cannot and does not have enough enough reliable data to make an educated guess that is anywhere close to accurate, and the results I see on my own sites and those of my clients bear that out.
[edited by: birdstuff at 1:04 am (utc) on Sep. 29, 2005]
The only real and effective form of smart pricing is advertiser control over where their ads run. The advertisers know (or will quickly figure out) what a click is worth to them from a given site/page. A Google algorithm does not.
The fact that so many sites that were hit hardest by smart pricing are doing exceptionally well now that the advertisers do have a bit of control sheds a lot of light on the smart pricing fiasco.
The recent increase in traffic is coming mostly from Google so it's only as good as what Google sends me ;)
It's been suggested that Google tracks referring URL, but I doubt it -- some traffic is just "worth" more, certainly traffic for different keywords, even if it all comes from one place.
I also agree with EFV that it's the long-term that counts, and smart pricing in its current form is likely to bite Google where it hurts after the competition is out of beta and publishers start looking seriously at other options.
Publishers will always go with what makes them the most money. Brand loyalty doesn't exist in this business. I've seen 50k a month affiliate accounts leave for a few hundred dollars.
Contrary to popular belief, smart pricing works for the publisher as well. It adds a certain value to AW accounts, and that attracts advertisers, which in turn adds value to AS accounts.
Contrary to popular belief, smart pricing works for the publisher as well. It adds a certain value to AW accounts, and that attracts advertisers, which in turn adds value to AS accounts.
The concept of smart pricing is great for high-quality publishers. In practice, a smart pricing scheme that is as ineffective and seemingly random as this one hurts everyone - publishers, advertisers, and ultimately Google.
Overall for our network of sites we have dropped about 17.5% in ctr and roughly 36.2% in epc. When this is over tens of thousands of verticals and hundreds of thousands of clicks per month you have a good picture of whats going on. I have spoken to quite a few of my buddies who are also in our range of revenue/traffic and they are seeing similar drops.
This started as far as we can tell immediatly after the following 2 events.
1. the change in adwords bidding at the end of last month
2. when G started messing with ad units last month, ie leaderboard showing 1 ad blahblahblah.
Does G drive us insane, yes but they are still the best game in town for now.
Just keep riding that gravy train until it derails:)
The concept of smart pricing is great for high-quality publishers. In practice, a smart pricing scheme that is as ineffective and seemingly random as this one hurts everyone - publishers, advertisers, and ultimately Google.
You can't call smart pricing random. It's a sequence of code that adjusts the price of clicks on the fly. There's no lack of order nor any point of chance. Just because a publisher doesn't like the outcome doesn't mean the system is flawed. How is it ineffective? The only people I've heard complain about smart pricing are publishers, who traditionally have an inflated sense of click value. You get paid what the click is worth, it's as simple as that.
smart pricing ... adjusts the price of clicks on the fly
On the fly? I don't think I heard Smart Pricing desacribed as an "on the fly" process before. I've thought it more like a periodic event, where a page is assigned a value factor and any click from that page adjusted in cost accordingly.
Could you describe what you mean by "on the fly" as it applies to Smart Pricing?
more people need to realise that Smartpricing is very, very limited in what it can do - and doesn't work in industries where conversions aren't the desired outcome and/or can't be tracked.
Let's not forget that "conversions" don't have to be e-commerce transactions. With Google's conversion tracking, a "conversion" can be any kind of "business action" that the advertiser desires--i.e., anything from a sale to a registration to an inquiry to reading a certain number of pages on the advertiser's site.
Also, AdSense CPC ads are meant to be direct-response ads (not branding ads), so adjusting prices downward for clicks that are statistically unlikely to result in "business actions" isn't unreasonable.
Finally, "smart pricing" doesn't have to be perfect. It merely needs to work well enough for advertisers to feel they're getting better value with smart pricing than without. The proof is in the pudding, and you'll see pudding aplenty if you look at Google's quarterly earnings reports.
(BTW, kudos to ncreegan for the most succinct and sensible defense of smart pricing that I've seen in this forum!)
The problem is that it is largely arbitary at the moment. There is no suggestion from Google that some industry sectors (particularly where conversion/action tracking is largely impossible) are not subject to Smartpricing. Further, any system which relies on extrapolation to the extent Smartpricing does has a major flaw when sample sizes are small. (Bear in mind that not all advertisers allow Google tracking, and that, even among those that do, not all actions are tracked. Some sectors/websites/advertisers may have too small a sample base for accurate Smartpricing calculations).
<speculation>
And, I think I've just been able to leverage this to my advantage. One of my sites talks about a product, widgetA, that people do buy online and conversions are easily tracked. I've been filtering traffic to those pages and seeing the WidgetA channel showing much higher CTRs. The strange thing is that other <high value> products I discuss elsewhere on the site - which can't normally be conversion tracked are showing a much higher EPC since I introduced that filtering to the WidgetA pages. I can only attribute it to the larger conversions advertisers are seeing from my WidgetA pages. The eCPM for WidgetB has been rising and is now well over $100 and about 250% of what it was before.
</speculation>
Could you describe what you mean by "on the fly" as it applies to Smart Pricing?
I was speaking of the analysis and output of a pricing adjustment based on input, not the life of a click. However, I would find it strange, given Google's affinity for parallel data processing, that the adjustment would not be real time.
From a logic perspective, I doubt a page would be assigned a universal smart pricing adjustment. There are certainly situations where clicks from the same page would be more or less likely to result in an action based on the advertiser.
edited to remove something I wasn't sure about, and added part of a brainstorm at the office...
You can't call smart pricing random. It's a sequence of code that adjusts the price of clicks on the fly. There's no lack of order nor any point of chance.
You're right. I'm sure that the specific process that Google uses is not random at all. But the inherent flaws in that process (as stated many times before) result in smart pricing "discounts" that are indeed seemingly random. To revisit an old programming cliche...garbage in - garbage out.
But the inherent flaws in that process (as stated many times before) result in smart pricing "discounts" that are indeed seemingly random.
But are advertisers complaining?
As far as I can see, the only people who are seriously unhappy with smart pricing are publishers who have been hurt by it. And they'd be complaining even if Google offered scientific proof that smart pricing was 100% accurate and fair.
But are advertisers complaining?
Of course not. Why should they? Smart pricing can only help them, not hurt them regardless of how poorly it works.
...how poorly it works.
You continually elude to smart pricing flaws. I'll bite. How doesn't it work?
As has been said in other threads ad infinitum conversion should not be a pricing element. The role of the publisher is to provide a platform for our content and the reasonable display of ads. Once a user clicks on an ad, it is out of our control and we should be neither rewarded nor punished for any subsequent user behavior. Period.
Of course not. Why should they? Smart pricing can only help them, not hurt them
Precisely. And who are Google's paying customers? Advertisers.
A couple of other things to keep in mind:
- Smart pricing lowers Google's revenue per click--not just the publisher's. So it's in Google's own interest not to give discounts to advertisers except when they're warranted.
- Unhappy advertisers leave, but most unhappy publishers just complain. In the rare cases where unhappy publishers actually bail out on AdSense because they think their clicks are undervalued, Google isn't losing much since the departing publishers are likely to be less profitable than the contented publishers who haven't been hurt by smart pricing.
Bottom line: If smart pricing is keeping advertisers happy and isn't hurting Google, it isn't likely to go away soon.
Smart pricing, whatever it is, is supposed to adjust based on conversion data. Since that conversion data is supplied by advertisers (merchants), and it is a small sample, and it is in the merchant's best interest to underreport conversions, it is inherently flawed from the start.
Google has never stated nor hinted that smart pricing is based on advertiser specific conversion data. It would be folly to value clicks in such a manner -- Google would be cutting into their own profits. Adjusting the price of the same click between advertisers is not good business; it doesn't encourage competition.
Publishers cry "flaw!" based on a theory that just doesn't stand to reason.
As has been said in other threads ad infinitum conversion should not be a pricing element.
Earlier I wrote that the same clicks should not be priced differently between advertisers. Don't confuse this with what smart pricing really does -- discount clicks that are not as desirable as those delivered at the maximum bid. It's a standard of marketing that action oriented advertising is only worth the profits produced.
AdWords offers two main sources of traffic, the Search Network and the Content Network. The content network has a huge array of sites of different calibers and foci. Because of that mix and the nature of the delivery, as a sum, the value of the traffic delivered from the Content Network is going to be lower than the Search Network1.
By charging Search Network price for a click that is of lower value, Google would lose advertisers. By standardizing click price based on an average, Google would lose potential revenue.
Finally -- why not charge a standard discount for Content Network advertising, and eliminate Smart Pricing? It comes down to revenue, again. By adjusting price by the factors they use, they stand to maximize both earnings for themselves and quality2 publishers and value to advertisers at the same time.
1 - Furthermore, most anyone with enough internet marketing and or affiliate marketing experience will agree that on a whole, search traffic converts better per click than content traffic.
2 - I've used quality here to describe publishers that are consistently able to send traffic that results in actions.
By charging Search Network price for a click that is of lower value, Google would lose advertisers.
I wonder how many publishers would support the other obvious choice, which would be to purge the network of all publishers whose clicks don't convert at search or near-search rates?
I wonder how many publishers would support the other obvious choice, which would be to purge the network of all publishers whose clicks don't convert at search or near-search rates?
This assumes that Google can make such a determination. By now it's quite obvious they can't - at least not with any reasonable level of accuracy.
And while the concept of smart-pricing gives many publishers a warm-fuzzy feeling, a smart-pricing that doesn't work as intended hurts them as well as publishers by driving many high-quality publishers to abandon running AdSense on their site(s). But ignorance is bliss as they say.
My figures may not impress many, but they are true and they are mine. Here's what Adsense did for me today, on a day that I knew would be solid traffic on a couple of particular pages:
First 1000 pageviews: 39 clicks, 6X earnings.
Next 1000 pageviews: 16 clicks, < 1X earnings.
I can understand some fluctuation, but clicks 2.5 x lower and earnings 6 x lower for the same number of pageviews falls well outside the range of statistical probability and is nothing but a scam, plain and simple.
I've criticized Google a great deal in the past, but today's fiasco is nothing short of robbery. While I'd love to leave them completely, I cannot right now. I am forced to stay on with them because even though they provide a highly questionable system of revenue, they still pay more than any other pure ad system.
I don't trust them. I don't like them and I will work to change them into an ad system that WORKS for publishers and advertisers alike.
And while the concept of smart-pricing gives many publishers a warm-fuzzy feeling, a smart-pricing that doesn't work as intended hurts them as well as publishers by driving many high-quality publishers to abandon running AdSense on their site(s).
1) The concept of smart pricing is to give advertisers a warm, fuzzy feeling.
2) Is there any evidence to support the idea that Google is losing "many high-quality publishers" because of smart pricing?
