Forum Moderators: martinibuster
According to PhoCusWright, research suggests that behavioral ads convert consumers at a higher rate than contextual ads do. Yahoo and MSN are already offering behavioral targeting, and "Google has filed related patent applications and is likely to weigh in in the near future."
In the PPC space, the holy grail would seem to be contextual targeting combined with behavioral targeting. (Think of Smart Pricing, but with ads served--not just priced--according to data obtained by tracking user behavior.)
If and when behavioral targeting becomes an important part of the AdSense ad-serving algorithm, audience quality could matter more than it does now, because sites that don't perform well for advertisers may receive few, if any, decent ads--and sites that draw large numbers of users who don't have the "intention" of buying a given type of product or service (as determined by observing their behavior on other sites) may also be at a disadvantage.
Good or bad for MFA's I wonder?
Probably good, for them. Theoretically, you could have a completely blank page, and the visitor would be served ads that are relevant to their surfing/buying/interest history.
PS. Amazon has behaviorally bases ads, now, too.
Consider the family computer: little Johnny looks for music downloads, games and the occasional swim suit model; mom looks for good deals on clothes, jewelry and cures to family ailments; and dad looks for hunting and fishing gear, car accesories and the occasional swim suit model.
How is the third party behavioral tracking program going to know who is on the computer at any given moment?
No matter how good they (behavior trackers) think they're targeting is, or the bells and whistles of their particular methodology, it's always going to be a shot in the dark.
Meanwhile, the website publisher can lose big (bad ad targeting) and maybe get a big fat EPC once in awhile when they get it right.
IMHO
Just yesterday I was looking for Halloween-Stuff for a big party in my house and I found an interesting ad on Google and clicked on it - I was SO amazed a few hours later when I found that exact ad on my site which is not at all about Halloween!
So I checked on another computer if I would find that ad again on my site - but there were completely different ads!
All Google needs to do is wake up to the fact that people who surf to rubbish timewaster sites, such as MySpace or YouTube or game/humour sites or picture galleries, are demonstrating from their behaviour that they aren't really interested in being sold anything at all.
It seems to me that surfers who choose to visit specific informational sites are already being targeted based on their behaviour.
That's somewhat true of users who visit niche sites, although even then, there's no assurance that every user is a prospect. For example, a visitor to a travel site about the Vatican might just want to see photos of the Pope or might be gathering information for a school report. With behavioral targeting based on user tracking, Google could make an educated assumption about whether the user of that Vatican travel site was likely to be a prospect.
As for the question of whether behavorial targeting works, that really isn't the issue here. If advertisers think it works, and if Google and other ad networks make ad-serving or pricing decisions based on assumptions about user behavior, then publishers will be affected (for the better or worse) whether behavioral targeting is based on hard science, crackpot pseudo-science, or mere guesswork.
Ann
It might sound like a good idea, and it is something that maybe could be sold to advertisers BUT, (and it a whopping, huge "but") Google would have to deliver on it.
Before they implement this, the technology would have to be in place to do the job, and it would have to work. The problem at the moment is that in it's current state, Adsense isn't capable of targeting the stock of ads it already has to appropriate sites a lot of the time. I don't think anyone can see them actually being able to make this work in practice bearing in mind they can't make what they have been working on for the last few years work. In fact, all these "Upgrades" are simply making things worse.
I think we are talking a very long time in the future - if ever.
I think we are talking a very long time in the future - if ever.
Maybe google is impaired, but maybe you should explain to amazon why this kind of targeting won't be coming soon. The fact they've been doing it for months now...eh...small detail.
Sounds like vapour-ware to me
Hmm, Amazon's Omakase seems to do exactly this. Technically it's in beta, but it combines: context sensitivity, prior customer interest and prior site success to deliver an ad to the visitor.
Due to that artial context sensitivity it's a problem compining it with the adsense ToS though (even if they don't really compete directly, and even if Amazon's ToS specify they have no problem with adsense on the same page)
I wish amazon would leave out the context sensitivity and just use the site's prior successes in selling and the customer's past interest. But no such luck so far.
Maybe Amazon has made it work - maybe not. The difficulty is that adsense / adwords is huge, ancient bloatware that has been patched so many time that Google probably don't really know what it does any longer. I don't honestly see Google being able to bolt anything else onto the thing and make it work. They don't know what QS/smartpricing/targeting (who all work separately and in contradiction of each other) will do given any particular circumstance - throwing more data at the think might simply confuse it even more than it already is. It's certainly not going to give the predicted result.
If we are talking Google here, the technology isn't there to make this work, and as they aren't selling it, I suspect they haven't rolled it out if they are working on it.
One can argue about how well behavioral targeting works or doesn't work, but it's already being used by Google's main competitors (not just by Amazon.com).
I suspect that behavioral targeting's real value is on portals, newspaper sites, etc. where audiences aren't defined by what they're looking at (as they presumably are, at least to some degree, on highly focused niche sites).
There's no reason why AdSense couldn't have both: Section targeting to help Google figure out that a page is about widgets, and behavioral targeting to figure out that Joe User is interested in widgets.
Let's assume that contextual targeting of ads is working, and to be honest apart from MFA's and some totally surreal targeting it's pretty good much of the time. Let's assume that the ads shown are based 50/50 on contextual and behavioral patterns.
The behavioral patterns is where I feel there are going to be too many variables to allow it to work properly, and if it doesn't work properly it's going to work against advertisers, and us.
OK - take our household. Two adults, two kids aged 9 and 11, four computers networked. My wife spends most of her time online either at money saving forums or doing geneaolgy research. I spend most of my time online researching for my site, here and counting the Goolgebucks (with dismay lately :( ), my kids have a mixture of chatting with their buddies, some online games that we approve of and some homework research. Now our patterns of use that could be used here are quite distinct. The difficulty is that (apart from my wife) the rest of us use whatever computer is nearest and who has won the fight for the comfy chair in the study. It's mine by rights, but if I claim that right, I then have to sit next to a 9 year old playing a noisy game, so I take the laptop into the toilet for some peace - assuming my wife hasn't had the same idea.
So our individual browsing patterns are spread about over several computers, and I think to work well it has to be assumed that the computer has one main user. In some circumstances that may well be true, but in an equal number of cases it isn't. How many people have just one PC in the house that is used by all family members? Maybe the main buying is done by people surfing from work, in which case there is a high likelihood of correctly targeted behavioral ads in that case. What about people NOT surfing from work?
Back to the idea that 50% of the ad content is based on our browsing habits. As seen above, on 3 out of 4 computers in our house there is a very mixed bag of browsing, and working out ads to show relevant to my buying habits (and I'M the one with the credit card) is not going to be easy. In fact, in our household I can see me getting ads for absolutely nothing relevant to my buying habits.
So take the logic to another level. We all know that Adsense works better than any other ad serving program because the ads are relevant to the site. I've tried a few of the so-called competitors, and the ads are absolutely abysmally targeted, and the returns are shocking. Google works for both advertisers and publishers because of the tightness of the ads to the content. If I now visit my site, assuming the 50/50 split I might see two ads for relevant services and two ads for stuff related to the main user of this PC - my 9 year old son. Even the MFA's (though I hate them) could be considered more relevant.
I don't think that scenario is going to work for advertisers, and nor do I think it's unusual either.
Looking at it from the publishers side, the message is one of "it doesn't matter how good your content or visitorss are, we are going to show lots of ads that aren't relevant to the site or visitors". As we all know, relevancy of ads is key to our (and google's) fortunes.
So nobody knows if Google are using this already, or might introduce it. I personally think that whereas contextual works reasonably well (lot's of room for improvement of course), behavioral makes too many assumptions. If it assumes each PC has one main user with a credit card, then it might just work - problem is with that assumption. I don't think it's very sound. Let alone the fact that actual buying habits may be very different from browsing habits.
Looking at it from the publishers side, the message is one of "it doesn't matter how good your content or visitorss are, we are going to show lots of ads that aren't relevant to the site or visitors". As we all know, relevancy of ads is key to our (and google's) fortunes.
Not necessarily. It could depend on the publisher.
On a newspaper site, for example, behavioral targeting could make a lot of sense, because the person who's reading a story on roadside bombings in Iraq probably isn't the market for related products or services, but he might conceivably respond to an ad for things that the ad network knows he's interested in.
Even on a niche site, behavioral targeting could be useful. Let's say that I'm a user who's visiting a site for photographers, and I'm browsing a gallery of photo-contest winners. I'm probably not going to be in the market for stock images or whatever other ads normally might be associated with a photo gallery, but if Google knows that I've spent a lot of time looking at reviews or ads for Canon cameras, it might be able to catch my attention with an ad for Canon's newest DSLR.
Side note: Topix.net, a news aggregator, has been combining behavioral targeting (based on readers' habits) with AdSense ads for quite some time. See the thread at:
[webmasterworld.com...]
As I said in the earlier post, assuming the person who is doing the browsing on a particular computer is the person with the credit card, then showing ads targeted to their interests makes a lot of sense. Problem is that it's a gamble for the ad server. They have to second guess what they actually spend their online cash on, and they have to make the assumption that the person is the main user of that particular PC. If that was the case in the majority of cases, then that's a gamble worth taking. But seeing how diverse the internet users are, I wonder of it's too much of a gamble to take. especially as adsense only works well because of ads that are targeted to site content.
But seeing how diverse the internet users are, I wonder of it's too much of a gamble to take. especially as adsense only works well because of ads that are targeted to site content.
Or page content, and that's the rub. I think behavioral targeting will be of great value on newspaper, portal, and other general-interest sites where audiences aren't necessarily researching purchases related to page content. Just now, I looked at a WASHINGTON POST article headed "McCain jokes about suicide if Democrats win Senate," and all of the Ads by Google were for for ISPs. Those generic ads probably don't do very well, but if Google knew (for example) that I'd bought transatlantic airline tickets or that I'm a regular reader of the POST's travel section, it might be able to serve ads for online ticket consolidators, luggage, hotel chains, etc. that would interest me more than ads for "Get High Speed Internet" or "[dial-up ISP name] only $4.97 a month."
At any rate, behavioral targeting is already in use on the Web, and I'd guess that it will be used by AdSense at some point--if not across the board, then for premium publishers for whom contextual advertising in its present form doesn't work very well in many cases.
It does make sense for more general pages like newpapers.
:)
How much would I have to bid for a pair of eyeballs attached to a user who had just typed:
"Where can I buy blue widgets online?"
Ok, maybe thats just a phrase in AdWords.
I have to say that Amazon's behavioral targeting ads converted absolutely worst for me. Absolutely. Is there anything absolutely lower than 0%? Maybe if someone saw the Amazon ads and decided they wanted to return something they bought from Amazon through my site previously. That may have happened too.
