Google’s Inside Adwords blog announced a limited test of a new advertising system on Tuesday, which would allow advertisers determine the value of certain actions (pay-per-action, or ppc).
The AdWords Help Center has an even more extensively detailed explanation of what Pay-Per-Action is, and how it might work. in their Pay-Per-Action FAQ. In both, there’s discussion that under this kind of approach, the advertiser would determine what value a pay-per-action add might cost.
A new patent application from Google explores this concept of having advertisers define the value of a click, or impression, or action – such as a specifically defined conversion. This includes how much an advertiser might pay to appear on a specific site, or what they might be willing to pay for a certain number of impressions or clickthroughs for different keyword phrases. Pay-per-action is one part of this larger system, and the system could also include offline advertising options.
A long and detailed document, the patent application is worth a read if you engage in purchasing advertising through a search engine:
Flexible advertising system which allows advertisers with different value propositions to express such value propositions to the advertising system
Invented by Sumit Agarwal, Gregory Joseph Badros, and John Fu
US Patent Application 20070067215
Published March 22, 2007
Filed September 16, 2005
Abstract
Different advertisers that may have different value propositions and that may desire different types of ad spots may effectively participate in an advertising network. An advertiser may express various value propositions using various types of offers, such as offers (or maximum offers) per impression, selection, and/or conversion (or some other user event). Probabilities of events, associated with offers, occurring may be used to allow different advertisers with different value propositions to compete against one another in an arbitration such as an auction. Advertisers may target the serving of their ads to keywords (search and/or content), particular publications or properties, particular vertical categories, other types of ad spots, etc.
I just signed up to beta test this new service. I am very curious to see how Google plays in Pay-Per-Action, as I have grown weary on untrustworthy affiliate companies as of late (especially CJ – they drive me nuts!)
Good to hear that you’re trying it out, Eric.
It looks like it might provide a lot more flexibility for advertisers who want something other than just impressions or clicks.
I’ll have to catch up with you on how you like it.
What? What’s new about CPA? Anybody heard about Affiliate Marketing before? CPA is the 2nd most compensation method in affiliate marketing, right after CPS (Cost Per Sale) also known as Revenue Share (a Percentage of the Order Value, usually excluding Tax and Shipping cost, placed by a customer who was referred by an Affiliate/Publisher).
PPA is the same as CPA as is PPC and CPC (cost per click).
Bill, you read the patent already in greater detail, what is so special about Google’s method to justify a patent over it? Which part of CPA did Google enhance that is not already done by Affiliate Networks that had 10 years to improve and tweak it?
May be I should file a patent that describes how to get paid for helpful tips that will allow businesses to improve and become more cost effective. Let’s call it Tip Fee, or Consulting Fee, oh that one is taken already and describes how to get paid for helpful consulting that will allow businesses to improve and become more cost effective. 🙂
Since this is “HOME TURF” of Affiliate Marketing, following posts on Affiliate Marketing related blogs might make an interesting read for anybody who is not very familiar with Performance Marketing aka Revenue share or Cost per Action.
ReveNews
Google’s Is Now an Affiliate Marketing Network
CostPerNews
Google and CPA: “It Starts With an Earthquake, Birds and Snakes…â€
Cheers!
Thanks, Carsten.
Appreciate the links to some Affiliate marketing posts on Google and CPA.
The patent application is much broader than cost-per-action, though I thought that the timing of the announcement of Google’s CPA beta, and the publication of this patent filing were timely.
Instead of a system where advertisers bid against each other, the focus of this is more of an advertising network where advertisers publish what they would be willing to pay for impressions, or for clicks, or for actions, and possibly where they would like to see those published.
In the next to last section of the patent is an example, using two different advertisers who each have presented what they would be willing to pay for different types of advertisement (in the example, impressions and click-throughs) for different types of keywords, and in some specifically named premium locations (the web sites for “Sports Illustrated” or “Parenting”). The example shows how the ad network might choose one of them for some of the ad opportunities available, and the other for some of the other ad opportunities. I wish they had included an CPA type of ad in the example, but they didn’t.
There’s also some discussion of how this system could be used for off-line advertising, such as banners at stadiums.
Regardless, CPA isn’t really the focus of this patent, but it is major part of it as one of the types of advertising that would be included amongst the ones that advertisers could set values for.
CPA definitely isn’t new, Carsten. But an advertising system like the one described may be.
The costpernews.com article, and one of the comments under it seem to capture why a system like this is needed. From the comment:
Since an advertiser can specifically propose a price and site for placement under this system, maybe this is being built in, somewhat.
Hi Bill,
thanks for the comment
When you talk about “the Ad-Network decides” , Google is meant and “decides” is probably something algorithmic. I simply can’t visualize a office at the Googleplex, full of designers and advertising folks that move advertiser Ads around by hand on websites and decide what goes where.
This seems more to be about Google providing advertisers the ability to specify their spending limit via a compensation method they know, prefer or/and feel more comfortable with. Google does then the number crunching to compare each Ad with the other on the same basis, probably CPM, because that is the metric to where they can “convert” CPM (doh), CPC and CPA to
1000 impressions for $10
CPC Ad with 1% conversion rate and $1 bid per click translates also to $10 CPM
CPA Ad with 10% conversion to clicks and 1% conversion of clicks for the desired action the Advertiser is willing to pay $10 for also translates to $10 CPM.
Now based on those basic numbers is it possible for Google to create a similar ranking model as they already did for just ranking CPC Ads that allows them to determine which Ad performs better under which circumstances. They must track the click conversion of CPM ads for sure and make it part of the equation or they would be off in some cases. If their technology requires to do the same “simple” and “unified” implementation on the Advertisers side, regardless of the compensation methods used by the Advertisers, Google could even track the CPA as well without the Advertiser even realizing that.
Maybe are they just fine with what they already got, the few advertisers that use their conversion tracker, Google analytics, Google checkout , Google toolbar, Google personalized etc. For the few cases where they can’t track the conversion directly or indirectly will an average of comparable Ads and markets be sufficient I guess.
I am just speculating here. hehe
You know what’s funny though. I had last June until about September some email communication with an interesting fella who is a Mathematics Professor who is going a bit into marketing. He found me on the internet somehow and contacted me, because he thought that I might be able to help him a little to understand affiliate marketing.
Very interesting conversations.
His specialty is Bayesian Inference which is already a popular among serious PPC search marketers and also among site conversion testing and optimization companies that are able to determine with pretty high accuracy which version of a page converts best, without the need to test all the millions of combinations that might be possible, but impossible to test via classic split or even Multivariate testing.
He wanted to apply the same mathematical principles to determine (as an affiliate, so the same Google might wants to do, but from the other side of the “fence”), which Ad to choose for each location on the website. It would basically be able to determine which compensation method and amount of commission (CPM, CPC, CPA and CPS offers, even combinations of them) is the most profitable.
Maybe he was onto something. This thing was the first that popped into my mind after I read your comments. My first thoughts are usually the right ones.
mmmh.. does this spill of thoughts trigger something with you?
Will be interesting to see unfold, one way or another, so much is sure.
Hi Carsten,
Right, the decision as to which ad to show is algorithmic, and your example of a method to compare may be similar to what Google would try to do.
I’m not sure how good Google is at estimating a CPA when they can’t track conversions directly. I was discussing that with someone yesterday, and they weren’t impressed with the numbers that Google was showing them, which were estimates.
I wonder sometimes how much of conversion testing and optimization is a science, and how much of it is an art. If you’re familiar with William Gibson’s book Pattern Recognition, and the main character just had a knack for understanding which fashions would become popular with unfailing intuition. I wonder sometimes if some people are capable of making the right decisions, and asking the right questions about what to test, without going through an incredible number of testing cycles.
Having said that, what I like about this method is that it appears to provide people with more flexibility, and more room to experiment, and try different things.
Nice analysis, Bill. I would not have guessed that Google was that far out in front 🙂
re: Pattern Recognition – I agree with your sentiment, Bill.
Those without ‘the knack’ can likely do it eventually with science, piecing together a changing picture in multiple dimensions, but at what cost?
So many variables right now, without some intuition, it would seem like we are doomed by crude tools…
Re: CPA(ction) Taking this concept to it’s logical extreme, this would mean Google takes over my marketing and I sit back and answer the phone… Not going to happen IMHO. Anyone making money with CPA will be driven to make more with CPC or other fundamental advertising.
-Jozian
Thanks, Jeff.
I’m not sure that Google wants to take the responsibility for marketing that far, at least not for most people. It might be too labor intensive for them to do that, at least upon a large scale, and we’ve seen them express an interest in automating as much as possible.
Besides, there’s television, radio, print, display ads, and other areas for them to tackle. The challenge there is in making it easier for people to advertise everywhere, with values that the advertisers determine. It’s in their best interest to be the vehicle for that movement rather than the ones determining the messages.
I’ve been using Pay per action for one of my ebooks I wrote, and have to say I am very satisfied with the results. Google needs to change a few things, but overall the PPA program is on the right path.
John
I would be pretty suprised if this gets traction on a wide scale. Google’s control relating to ad costs is fairly strong and I dont see them allowing advertisers to have more control. Even in this down economy, paid search seems to get stronger in terms of volume and macro trends. What really needs to happen is that webmasters need to become more sophisticated relating to multivariate testing and get higher conversions on flat traffic.
Hi Craig,
Testing, testing, testing….
I think that’s a major key to success – being willing to test, and then being able to understand the results of those tests. Even simple changes can make big differences sometimes.
The pay per action beta program that was the subject of this post was phased out by Google after their purchase of DoubleClick (see: No longer available), and two alternatives have been provided: Conversion Optimizer and Performics Affiliate.
Regardless of the value or availability of those programs, I agree with you about testing.
I am wondering if there are any industry averages for click-to-purchase rates for selling software on the Internet if you are paying for click throughs from Google. DoubleClick says that the click-to-purchase rates for email campaigns is about 4.2%. What would it be for AdWords?
Hi Mike,
Not sure, but an interesting topic. I tend to focus more upon news and information about organic search rather than paid search, but it would be really interesting if Google would share that kind of data.