Smart ad placement as an alternative to smart pricing
| 1:48 am on Jul 19, 2006 (gmt 0)|
In the past, we've talked about the possibility of an "AdSense Select" or "AdSense Gold" tier within the AdSense network, with sites being certified for quality and offered to advertisers at premium rates. Google hasn't introduced such a tier (at least, not yet), and one could argue that Google probably won't because it prefers to use scalable algorithmic solutions whenever it can.
Another idea, site-targeted CPC ads, might be more viable, since advertisers would pick sites for targeting as some currently do with CPM ads, but it may also be vulnerable to the "algorithm preferred" objection.
So let's look at a third possibility: an option (let's call it "Smart Select") that permits advertisers to run their ads only on sites or pages that that are deemed likely to convert for advertisers.
In effect, Smart Select would be a different incarnation of Smart Pricing, using the same underlying assumptions and technology but offering advertisers control instead of discounts. In other words:
With Smart Pricing: An advertiser currently accepts run-of-network placement in return for the promise of discounts on what Google considers to be low-quality clicks.
With Smart Select: An advertiser would choose a minimum "quality tier" (e.g., a percentile tier based on Smart Pricing conversion data), and ads would run only on sites that exceeded that minimum. Pricing would be tied to the quality tier chosen by the advertiser.
Now for the pros and cons:
PRO: Unlike manual methods (such as site targeting), Smart Select would be algorithm-based. This would make it scalable, would require no extra work for the advertiser, and would ensure that quality mom-and-pop sites wouldn't lose high-paying ads.
CON: Unlike manual methods (such as site targeting), Smart Select would be algorithm-based. This would evoke fear and suspicion among publishers and advertisers who don't trust Google's Smart Pricing technology.
| 2:18 am on Jul 19, 2006 (gmt 0)|
I actually like the sound of this. But I'm also one of those who believes that Adsense works best on sites with items to sell.
My site offers nothing for sale, and as a result -- advertiser conversions do not usually fair well.
| 4:10 pm on Jul 20, 2006 (gmt 0)|
Smart Select: Agreed on your points but how does this help if google is going to introduce CPA model(cost per action).I read that it already introduced CPA.
How and why should google implement smart select? Can you elaborate :)
| 4:25 pm on Jul 20, 2006 (gmt 0)|
|Smart Select: Agreed on your points but how does this help if google is going to introduce CPA model(cost per action).I read that it already introduced CPA. |
If Google introduces CPA (beyond the current beta test), it will almost certainly be an option, not a replacement for CPC, for reasons that have been discussed in this forum's existing "Google testing cost per action" thread at:
|How and why should google implement smart select? Can you elaborate :) |
How? They've already got "Smart Pricing" software that uses various types of data to predict conversion rates for a given account, site, and/or page. (We don't know exactly how it works, and Google isn't telling.) All they need to do is use Smart Pricing data for sorting, not just for pricing.
Why? Because some (probably many) advertisers would like more control over where their ads run, especially at a time when there are so many made-for-Adsense junk sites. Smart Select would satisfy that desire and make AdSense more attractive to advertisers who have been unwilling to risk money on the content network.
| 4:27 pm on Jul 20, 2006 (gmt 0)|
|un their ads only on sites or pages that that are deemed likely to convert |
I like the other 2 suggestions, but isn't Smart Select already there with price threshold and smartpricing? Sites that convert are priced higher than sites that don't, advertiser sets his max bid, and in effect his ads display only in sites that the "algo" deemed converting well, right?
| 5:59 pm on Jul 20, 2006 (gmt 0)|
I don't really want to discuss the technology aspect, as I think your intention here is to discuss an idea. Let's just assume it's technically workable for the sake of discussion.
As a publisher, I'm not sure what's in it for me. Clearly the advertiser will pay more. Is there an assumption to be made here that publishers will also recieve an enhanced pay out? That being the case, I can't really see that there is anything to dislike about the idea from the publishers point of view.
I think the idea will be pretty easy to sell to publishers, but advertisers may be a different kettle of fish. If I put my advertiser's hat on, I have specific requirements. I basically need to know that the premium I pay is worth the money, and it does indeed work for me.
For example, I would be happy to pay a premium for a quality conversion. To me, this would be a conversion of my choosing. Therefore, if Google were to be the one deciding what did, and what didn't convert it would not be for me. I want to choose what actions do, or don't convert. I would not pay a premium and then have someone else make that decision.
I would also be happy to pay more for non-converting placement, and clicks from relevant premium sites. Although I can see that a scalable solution might possibly do a lot of the drudge work, I would still require to have the final say in where my premium bucks were spent.
I do like the idea of a premium publisher tier attracting premium advertisers. It's really what adsense started out to be, as opposed to what it's become. As a publisher I can't see any reason to be opposed to the idea as long as I was recieving a premium price per click. As an advertiser, I think I would be looking for Google to make selections, and then allow me the final choice.
| 6:11 pm on Jul 20, 2006 (gmt 0)|
>premium publisher tier attracting premium advertisers
amen to that David, assuming all parties are angels all should go well.
| 6:19 pm on Jul 20, 2006 (gmt 0)|
|As an advertiser, I think I would be looking for Google to make selections, and then allow me the final choice. |
That's not a bad idea, though one could argue that (in most cases) a combination of Smart Select and the existing domain filter would provide that selection and choice.
| 6:37 pm on Jul 20, 2006 (gmt 0)|
I think for an advertiser who is selling a product it's a no-brainer to opt in to an enhanced tier of contextual publishers.
Scenarios and assumptions
1: There will be a cash drain from the normal contextual pool if you allow advertisers the ability to exclusively opt-in to the Smart Select publishers. Thus...
2: This would have to be an add-on to inclusion within the normal contextual network.
3: It would probably be called, "Contextual Optimization." To do it in a manner that explicitly means opting in to a higher quality tier is to imply that the rest of the network is crap.
| 7:04 pm on Jul 20, 2006 (gmt 0)|
|2: This would have to be an add-on to inclusion within the normal contextual network. |
Yes, the advertiser would be opting into a subset of the overall network that has been identified through Smart Pricing or similar data.
|3: It would probably be called, "Contextual Optimization." To do it in a manner that explicitly means opting in to a higher quality tier is to imply that the rest of the network is crap. |
There's nothing new about buying a subset of an audience as opposed to "run of network." Some advertisers would obviously prefer a higher-quality tier, while others might get a better return on investment with cheaper but less behaviorally-targeted RON advertising.
Also, AdSense is direct-response advertising, and in the direct-marketing world, buyers have had a spectrum of choices from "pretty much everybody" to "Tulsa funeral directors" to "owners of $2,000,000+ homes in the 90210 ZIP code who subscribe to THE ROBB REPORT" for a long time. Still, "Contextual Optimization" is a nice phrase, and it works for me. :-)
| 12:53 am on Jul 21, 2006 (gmt 0)|
Thinking about this some more, in many ways I'd expect Google to be doing much of this already. I'm not convinced that it doesn't already happen either.
As an advertiser, what I'd expect Google to already be doing is to match my ads to relevant content, and I'd expect them to be placing my ads on the sites Google knew converted best for my keywords. As it's an auction process, I'd therefore expect that if I increased my bid, my placement would improve. Obviously in search that means at the top of the search pages for my keywords, and in content that would mean (to me) that my ads would appear on the sites towards the top of the search pages, or sites that Google knew would convert for me.
My expectations of the current system as regards content is that if I bid 5c per click, then I would expect my ads to appear primarily on less popular sites. If I bid $5 per click in content, then I'd expect my ads to appear on many sites, but only pay the top whack on the top sites.
I think this is broadly what you are talking about happening, and IMHO it already does happen. My site is usually top 5 of Google for most relevant search terms. Therefore it's logical that for advertisers advertising to the niche it's prime real estate. Even at it's worst, I can't complain about the epc and at the moment since QS I'm averaging about $1 per click, and some channels reporting higher than that. So maybe much of what you suggest already happens. I'm sure that many of the sites in the sector do not get that amount per click - even after QS.
Therefore, I wonder what would sell the idea of paying even more to advertise. As far as I'm concerned I expect an increased bid to give me prime placement, and to pay a premium for sites that work, and have the ones that don't discounted. That is Google's current offering. Although I support the idea of a top tier, I don't really know what would sell me the idea of paying over and above what the current proposition is.
| 1:07 am on Jul 21, 2006 (gmt 0)|
|As an advertiser, what I'd expect Google to already be doing is to match my ads to relevant content, and I'd expect them to be placing my ads on the sites Google knew converted best for my keywords... |
I think that's a pretty big assumption. If it's true, then I wonder why Google hasn't publicized the fact that higher bids will ensure placement on the best-converting sites. To me, that would seem to be a more compelling message than "we'll give you smart-pricing discounts on sites that convert poorly."
| 1:47 am on Jul 21, 2006 (gmt 0)|
I think both messages would be equally important to an advertiser. But if you were an advertiser currently bidding at a premium in content, then what would you reasonably expect for the extra money?
There are advertisers that DO bid high in content right now. I think they only do that for one reason - it works for them. Therefore they must feel that the extra money they spend in content gets them something, otherwise they wouldn't be bidding high. Although I can't see anything in adwords help at a quick glance, I'd suggest that advertisers could reasonably expect to get better positioning in search AND content for paying a premium. And judging by what I see on my site, I think it does happen even though the help file doesn't specifically state this. I'd suggest that it's a reasoable expectation for an advertiser, and as Google is certainly more clued up on advertisers requirements than ours, I think it's likely that their expectations have to be met in order for them to place high content bids.
| 4:00 am on Jul 21, 2006 (gmt 0)|
My concern is not so much that I want a Smart Select type program to make more money. My real worry is that many advertisers are opting out of content. I would hope a tier including only quality sites would be something more advertisers would select.
| 7:22 am on Jul 21, 2006 (gmt 0)|
EFV: Again it comes back to options, the advertiser having option of selecting contextual network sites, there is site targetting in place now, he can choose sites to target however this is quite a cumbersome process.
Normally given google's approach of optimization and algorithms, smart select might be an option they might already have implemented in algorithms , right now our argument is that advertiser needs that option,basically when contextual started,the advertiser had no choice of choosing contentual network it was default option, then the option appeared which they can choose, later site targetting appeared for advertisers, now you are taking about smart select option,
I guess they will most likely come up with some sort of conversion based selection criteria(with pricing advantage), like how google now gives importance to theme based sites for ranking in search listings.
Adsense network is so huge and gigantic network that even if google decides to make some changes to it and tries to make a tier of out the network,It has to be carefull like martini pointed out,you cant make tiers since that would basically mean the other side of the network is crap.Given google's culture of not having hierarchy I doubt they wil go for tier based system.Instead more fine tuning of ECPM might happen to control quality, ECPM is a lever in google's hands which directly controls the advertisers spend, publishers earnings and google's profits.
| 2:56 pm on Jul 21, 2006 (gmt 0)|
|iven google's culture of not having hierarchy I doubt they wil go for tier based system |
Google has already gone in the direction of hierarchy by introducing separate bidding for the search and content networks.
Also, Smart Select wouldn't necessarily have to be configured as I've described it (e.g., as a group of percentile "tiers"). It could just as well be configured with a sliding scale--let's steal a word from Martinibuster and call it an "optimization filter" that would let advertisers choose the level of performance and Smart Pricing discount by with a slider control. For discussion's sake, we'll call this "Smart Pricing Select." You'd have:
(lowest performance / biggest discount ) 1 2 3 4 5 6 7 8 9 10 (highest performance / no discount).
With "Smart Pricing Select," an advertiser would have greater control over both audience quality and price--and by testing Smart Pricing Select at different discount/filtration levels, the advertiser would be able to find the level with the optimum ROI.
Side note: Earlier, someone wanted to know what was in this for publishers. The answer to that is simple: Smart Select would make the content network more attractive to advertisers who are leery of the current "you gotta take potluck" run-of-network approach, it would reward publishers whose sites are attractive to advertisers, and it would be better for small, low-profile publishers than the most obvious alternative (site targeting for CPC ads).