Has Contextual Advertising Peaked?

I think most of the first argument is false logic. A majority of the article is based around this quote:

"If search only represents 5% of online media time, it shouldn’t have 40% of the dollars, no matter how measurable search advertising is".

What it fails to mention is that adsense & YPN are displayed on the other sites that users are spending all their time.

Further more while CPA and lead generation are already big earners but not all publishers can get them to work so CPC is always going to be around. What I personally see happening is that advertisors will get better metrics, as mentioned in the article, but will use that to adjust their CPC budgets accordingly. This is already done automatically in a way via google smart pricing.

In any media don't certain segments merit a premium? The closer one is to the conversion the more value there is, right? At least I thought that was how it works with direct marketers placing classifieds, etc. What does it matter if relevant on topic ads seen by people actively trying to solve problem make up most of the market over ads by people just surfing around?

I and a couple friends noticed about a 20% decline in AdSense earnings over the last month across 3 different verticals. As a large AdSense ad buyer for a client in another vertical I also noticed about the same decrease.

I think with Yahoo!'s Publishing program being so topically irrelevant and slow Google is overly kicking in smart pricing to cut back their prices with the hopes of extending the depth of their contextual network advertiser pool before anyone else seriously enters the market.

Before Google brought commercial publishing to the webmaster masses, ad spend was tied up in networks, right? If that works so well, why did AdSense succeed?

Speaking of which, I tried YPN and it was so off target for so long it wasn't even amusing. It was on a site about music equipement that easily should of had advertisors of some sort or another but instead they pushed VONAGE and other unrelated stuff.

I hope they fix it.

Ohhhh, so advertisers are going to move to more "premium" sites in BURST (did they use premium and BURST in the same sentence?), rather than advertise on niche sites which are more relevant to their product.

The closer one is to the conversion the more value there is, right?

i think that's the big debate that scott karp and those who disagree with him are having. scott says:

First, just because you can’t easily measure the ROI of brand advertising like you can transactional search advertising doesn’t mean the value is not there. And advertisers know this.

brand advertising is further away from conversion and its value is not easily measured ROI style, but does that mean its less valuable? is it less valuable online than it is offline?

Before Google brought commercial publishing to the webmaster masses, ad spend was tied up in networks, right? If that works so well, why did AdSense succeed?

i think the answer to that question comes in one word: convenience. google made it super easy for anyone to put adsense on their web pages, and fairly easy for anyone to get started with contextual advertising. non-contextual advertising generally requires some graphic design -- that alone eliminates a lot of small businesses, or at least discourages people who otherwise would consider online advertising.

Warning: Long post. Strike that. Very long post.

Main point: Joanne is (partly) right, the rest is crap. This quote is from a source who knows as little about research as traditional media knows about SEO.
------------------------------------------------------------------

MSN is of course pointing their fingers at Google Adsense here. The fact (as I see it, personally) is that banner ads don't work, everybody knows it but nobody does anything because they're used to clickrates below half a percent. Banners are good for branding, and that's about it.

Of course, if you hit the right demographics, then the banners may be more relevant, and hence they just might work a wee bit better, but that's the natural development path for all advertising (including contextual) it's not specific to banners. Basically you only want to hit potential customers, hitting everyone else is a waste of money.

To move goods you need contextual (and contextual can be graphical as well). And, by the way, who says Google isn't investing in analytics?

And then to the core of the matter, "uniform metrics". For an example, just take a look at the Burst Media LLC "channels". Yes, the magnificent example from the article. That's a freakin' topical web directory, it has absolutely nothing to do with demographics or real people.

So, you look at that page and you see nice figures like, "88% male", or "43% $50K plus". Wow, "how neat" you think. But... just where does it say exactly how they got that information? Not on that page.

On another page I found this:

Burst collects voluntary survey data from the visitors on all of our member sites to help advertisers better target their audience across our network.

I bolded the informational part. So, they serve popups, right? Or, wait a minute, do they telephone these people and ask them questions? Send them an email with a link? Or, do they stop them on the street? Or do they collect that information during signup for purchase? For free membership?

Probably popups, I'd say. You all know about those, but if you don't one word is enough to describe them: Crap.

Even if it isn't popups, never, ever trust ANYTHING branded as "research" figures UNLESS you see every significant detail about the methods used - preferably up front, open, and in a language that YOU can understand. The method used is not only important, it is CRUCIAL in order to determine if the figures you're watching are worth anything. No method, no numbers - you might as well make them up.

For christ sake, we don't even know if it's the same people that has been asked about sex that have also been asked about income. We have no freaking clue how many were asked, how big a proportion of the user base that sample is, if they were asked during holiday season or distributed evenly across the year. We don't know how many of those asked did not answer. We don't even know if these figures are, say two years old, and we don't know if they answered the truth, or if they even understood the questions. But that's details.

We have less chance than an icecream in hell to know if the people that did answer looks in any kind of way like the people that did not answer, and we don't even know if they're the majority or not.

There are a lot of rules in analytics, and I'm not going to lecture here, I only need to state the most important one. You might know it from database programing, it's: garbage in, garbage out

Now, that was the method - ie. the words "collects ... survey data" from the quote above. Now let's look at the word "voluntary". Is a pop-up voluntary? Well, if you decide to input the required information without being forced to do so, one can probably say it's voluntary. So, if I pay you to tell me something, will that be voluntary? I guess so. If you need to fill in a form in order to subscribe to, or buy, something that you really want, would that be voluntary? Well, so-so, perhaps... If I tell you that you can win a great prize if you click this link and fill out a survey would that be voluntary? Well, I would think so. If you decide to answer a telephone survey? Well, yes. If you read about a survey somewhere and decide on your own to take it, just for the fun of it? Or because you're well-meaning in general? Or because you're a member of some special group (eg. activist, member of a political party, competitor, relative)?

The thing is, even if all these methods can be said to be "voluntary" to some extent, it just does not matter. The word voluntary sounds good, but it's not the right place to focus. Some of these voluntary methods yield garbage. The only right place to focus outside reliability is validity, which means that they should focus on measuring the things they say they are measuring.

Specifically, the respondents may be voluntary, but to heck with that if it's not a representative sample. A big sample does not make an accurate sample. And, as for big, 2,000 is a big number if it's the price of a cup of coffee in dollars but it's less than a fraction of a grain of sand in Sahara relative to web sites.

But, surely claus, dontcha think that a firm as large as BurstMediaLLC is reliable and state-of-the-art when it comes to surveys?

One word: NO. Not reliable. They might be state-of-the-art, but on the internet "state-of-the-art" is very low compared to other media, or other market research fields.

Never trust anyone who does not reveal their methods. I even have reservations towards the Nielsen//Netratings figures (well, basically they suck as well, for numerous reasons and purposes. Definitely not all, only most), and that's a pretty big research organization that do reveal part of their method when asked a whole lot of times (general well-polished headlines, not specific details).

So, with my background in traditional media research, whenever I hear some internet marketer talk about segmentation I have to shake my head. Usually they have no freaking clue what they're talking about.

(partly)

This, OTOH, is 100% true:

To join the elite, however, they will have to provide advertisers with reliable numbers. For this, many are already piling into associations that can vouch for a level of quality and provide uniform metrics.

I've worked with this for seven years now, and MSN has been part of that work force for a while as well (In this neck of the woods Yahoo was first to see the value of it though, they deserve that much credit). So, in this respect Joanne is absolutely right, it's the journalist that's comparing apples and oranges by confusing BurstMediaLLC's sales material with research.

PPC and banner ads are exercises in negative reinforcement. The relevance factor of paid ads are never as good as the real thing, and over time you train people not to click on them.

(Even with Google throwing in the relevance factor in the ranking formula, you're still only getting something, at best, 50% as good as it could be.)

That's why it's funny to see this article say that banner ads are going to make a comeback.

They might make a comeback, but they won't be any more effective.

when contextual advertising first came out, the argument was that it was actually useful to the end user. it wasnt distracting and it was actually related to the content on the page, which is what hte user was actually reading, and hence it was better for the user. or so the argument went.

now i think contextual advertising could get supplanted by behavioral advertising which will focus more on gathering information about the user rather than assessing the content of the web page (like contextual advertising does). contextual advertising is still advertising that is derived from attributes of the product (i.e. text on the page), while behavioral advertising is derived from attributes of the user. if this makes it more relevant for the user, presumably the publisher can charge more, and this could be one way the problem of lack of funding for good content gets solved.

if this is so i wonder do publishers need to focus more on not just producing good content but also on making their sites more interactive so that they can get more info out of users, and hence be in a better position to sell behavioral metrics to advertisers and ad networks. just as adsense-dependent publishers focus on content and keywords, perhaps publishers seeking premium ad revenue need to focus on getting behavioral metrics that can be resold once the behavioral ad market develops.

It is all a bit hard to take with AdSense ads running across that blog.

I have a problem with this article. The idea of valuing advertising based merely on time spent is very naive. It is also a hold over from traditional media.

Search should still command the lion share of profits (higher CPC) compared to other types of sites because search is very good at identifying the user's mode when they are on the web. When a user searches for something they are essentially opting to disclose they are interested in a certain subject.

There is a reason why bid amounts for search keywords are almost 10X higher than the same keywords on AdSense. Simply put the AdSense keywords just don't work as well as AdWords.

IMHO, the challenge that Google has to justify their valuation is to be able to create lots of inventory that creates contexts that are valued by advertisers. That is why I have a problem (from a monetization standpoint) with the slate of recent Google offerings such as GMail, Google Maps, Google Video, and Google Earth where they are having a hard time matching up context on those properties. I think Google's time is better spent developing vertical search engines in categories that have higher revenue potential (ie. fixing Froogle!).