| This 46 message thread spans 2 pages: < < 46 ( 1  ) || |
|Pay per click, soon to die?|
| 3:24 pm on Mar 28, 2006 (gmt 0)|
As most of you probably know, billing for adverts on major search engines occurs via a Cost Per Click model. Each click on a particular ad drives up the bill for advertisers.
I have been thinking about why this business model is unstable and likely to collapse in the near future; I would be interested in hearing your opinions on this matter.
As research shows, more and more businesses are turning to search engine paid marketing strategies for their promotionnal needs, as a result cost per clicks on average are rising due to the growth of advertisers. In the current state of things I would without a doubt say that most advertisers are obtaining good return of investements with this method of advertising. Fast forward to a year's time, the number of advertisers would have grown significantly along with the cost per click prices...
My feeling about this is that as more advertisers use the cost per click model, it will become tougher to break even and eventually will come a point where advertiser's return on investment will become null.
Add to this the growing problem of click fraud and I think you will find that the PPC/CPC model hasn't got a bright future...
Your opinions and arguments will be much appreciated;
| 8:13 pm on Apr 2, 2006 (gmt 0)|
Let me rephrase my last statement about click fraud, since I really meant to generalize it to all types of performance-oriented models (CPC, CPA, CPM where impressions are counted as ad fetches or views, etc.). All of these models are more susceptible to performance fraud (the ability to generate clicks, impressions, calls, or what have you) than the types of models that are present in other media (newspapers, TV, etc.) because of the relative ease with which traffic can be generated on the Internet.
For any media upon which chargeable actions can be cheaply and anonymously generated, there will be fraud that is difficult to stop. Charging fixed fees reduces both the ability, and the incentive, to commit performance-based fraud.
|That leaves you to make it more difficult. I assert that click fraud can be made sufficiently difficult that it no longer makes financial sense. |
Given current Internet architecture and technology, how? And how would this be done without eliminating nonconverting traffic that is currently charged for?
|This is relevant because I believe CPC to be the most efficient overall model for delivering customers to advertisers. |
Sure, if you subtract the cost of detecting fraud, compensating its victims, etc.
|In terms of fraud, CPM does nothing to diminish the incentive for fraud. Impression fraud seems to be a more difficult problem to stop since the search engine has less transactional data to scrub for say, an XML call for a given keyword than they do for a click on a .js based ad. Further more, scrubbing on the click is computationally less expensive, since clicks happen 10-20 times less frequently than do impressions. |
This is because charges are set based on the number of impressions. If charges are set based on the (statistical) number of individuals, there is no fraud that can result by generating more impressions. Certainly, it is possible to defraud the system by tampering with the sampling of the audience, or with the actual billing mechanism. But this is far more difficult.
| 6:26 pm on Apr 4, 2006 (gmt 0)|
|Given current Internet architecture and technology, how? And how would this be done without eliminating nonconverting traffic that is currently charged for? |
I'm having trouble understanding this question. It is possible to minimize the effects automated click fraud [snip] but there is a cost, as one has to 'act on suspicion' on *some perfectly good clicks.
It's a small downside though, because the advertisers get the clicks for free.
Is that what you mean by "eliminating nonconverting traffic"? As long as humans within a relevant geographical area are initiating searches, I don't think you can 'eliminate' the portion of said traffic that necessarily won't convert. I don't think you would suggest this, but that's another way your statement could be read, it seems.
[edited by: bakedjake at 8:37 pm (utc) on May 19, 2006]
[edit reason] shameless self promotion [/edit]
| 10:46 pm on Apr 4, 2006 (gmt 0)|
|I'm having trouble understanding this question. It is possible to minimize the effects automated click fraud - we do it on our network right now through a combination of client side and server side scripting - but there is a cost, as one has to 'act on suspicion' on *some perfectly good clicks. |
Sure, it is possible to minimize the effects of automated click fraud -- depending on the method of automation that's used, and how dissimilar the fraudulent traffic is to what you generally consider to be "legit" traffic. That's not what I was arguing. You claimed that it was possible to make it difficult enough that it no longer makes financial sense. That statement alone is highly doubtful, because click fraud takes place in locations where a few clicks per day can actually amount to a decent day's income. So it makes perfect financial sense for the fraudsters to engage in such, and all they need to do is coordinate amongst themselves how and when they will click. They don't even need to compromise any machines.
Now if you argue that such fraud amounts to a drop in the bucket to the advertiser, again, this isn't necessarily so.
|It's a small downside though, because the advertisers get the clicks for free. |
Is that what you mean by "eliminating nonconverting traffic"?
Yes, that's what I meant. Sure, it's favorable to the advertiser, but not so for the SE or ad network, which wants to maximize the number of clicks it can actually charge for.
When CPC first began, there was a feeling that the click was sufficiently "close to the sale" that advertisers would get ample conversions to justify their costs. This philosophy basically ignored the threat of click fraud, especially the types that are difficult to track, because they basically look like a casual visit to the advertiser's site.
| 12:29 am on Apr 5, 2006 (gmt 0)|
yeah click fraud is a big problem, you can only hope that google and yahoo go after enough people and change things enough to keep up with the fraudsters to make it hard to do
| 3:56 am on Apr 5, 2006 (gmt 0)|
|Sure, it is possible to minimize the effects of automated click fraud -- depending on the method of automation that's used, and how dissimilar the fraudulent traffic is to what you generally consider to be "legit" traffic. That's not what I was arguing. You claimed that it was possible to make it difficult enough that it no longer makes financial sense. |
You're making some assumptions about how effective fraud protection can be. While I could never argue that any fraud protection is unbeatable, I continue to contend that one can make fraud sufficiently expensive as to reduce it to a point of irrelevance.
|Sure, it's favorable to the advertiser, but not so for the SE or ad network, which wants to maximize the number of clicks it can actually charge for. |
Search engines are third party beneficiaries of a valuable transaction which they enable between party one, the consumer and party two, the advertiser. While it might appear that it's in their best interest to "maximize the number of clicks" they can charge for, I think it's clearly in their long-term interest to maximize value to their partners.
Not only does doing so decrease advertiser decay and churn, but it also maintains a robust bidding marketplace.
Charging advertisers for automated click fraud might appear tempting, and plenty have fallen prety to doing it, but it's a short, hot burn.
GB, I think we're on the same side here. You are probably right that a model based on statistical impressions would do a lot to curb click fraud, but I think that advertisers are comfortable with CPC and the model can and will work.
| 7:17 am on Apr 18, 2006 (gmt 0)|
Yes, and my company is already positioned for the change. We engaged in PPC obviously now, but I think it will move in the direction of a hybrid of CPA + CPC. My company is already closing deals on this model. Fixed low CPC + CPA pays for the space with a little profit and costs the advertiser a fraction of the PPC model. In compensation, CPA's will go up.
Also, we do something others avoid - we prefer to host our PPC ads only. This way we can control all traffic.
| 7:53 pm on Apr 24, 2006 (gmt 0)|
Off topic a bit, but another point- I personally think that PPC will die not because of prices, but because people will just stop clicking on the paid listings.
I know I always click on organic listings, never paid, and I work for a SEO company. Before I started working here I didn't trust any paid listings. Now that I know, I still don't click. Lots of people don't understand how PPC works.
I'm 23 years old, and NONE of my friends click on paid listings. I have a feeling people younger than me don't either. It's just going to die as the people who do click stop using the internet.
yeah, possibly far-fetched...but, i don't know, maybe not.
| 8:12 pm on Apr 24, 2006 (gmt 0)|
We as marketers are a poor focus group on this topic because we already know how it works and do not represent the general populace. The data that I've seen seems to indicate an opposite trend.
From conversation with other marketers and search engine folks the consensus seems to be that people are getting lazier. For example many users will type www.example.com into the search bar instead of the address bar, then they will click on a link, sometimes organic and sometimes paid (depending on which is easier, I guess). The same seems to be true for actual search queries too; many users click on the first satisfactory result they see, either paid or organic.
| 8:19 pm on Apr 24, 2006 (gmt 0)|
I don't buy that. Can you show me this data?
| 8:33 pm on Apr 24, 2006 (gmt 0)|
No, actually, I can't.
Not a very convincing response, I know, but the data itself is private. I encourage you to dig through your historical data to find trends that pertain to your industry. It may be that your target audience is indeed untrusting of paid advertisements.
| 2:50 pm on Apr 25, 2006 (gmt 0)|
I guess I'm not being specific to industry.
I just think as the younger people move into these different industries, they'll be less likely to click on PPC ads.
What I'm saying, of course, is very unscientific, just me asking around to my friends and things. Everyone I talk to my age or younger is very suspicious of the PPC ads, and I imagine it's this way across the country. People just aren't aware of what they really are.
| 3:04 pm on Apr 25, 2006 (gmt 0)|
Yesterday I found a programmer (who revealed to me the secrets of BASIC long time ago) writing an URL in the Google toolbar of his brand new MacTel while he was searching something related with the work his doing for me.
| 8:06 pm on Apr 25, 2006 (gmt 0)|
Personally, I rarely click on any ads (CPC or otherwise). Same for my friends; arguably, most of them are either old-school network software engineers or don't use the Internet much.
I have heard that Joe and Jane Surfer click on ads that they think are part of the serps, or that look as relevant as any of the serps. I have also heard that there is a growing bias against ads, as people are either burning out on them or feel they are deceptive.
| 6:39 pm on Apr 28, 2006 (gmt 0)|
The future of PPC will not be pay per click on an ad but ppc on any link. Eventually to find anything on the net whether it be an ad or an article, media, etc. virtually every click you make will be a charged click, ppc or otherwise.
Think about it, the net started out free, every click was free there where no ads, what percentage of clicks do you think are free on the net now? In time every free link will become a paid link.
All the net is turning into is interactive tv. Do you think there are any free ads on television? I don't think so.
| 7:26 pm on Apr 28, 2006 (gmt 0)|
| 8:44 pm on Apr 28, 2006 (gmt 0)|
I doubt there is enough advertising revenue out there to monetize anywhere close to every link.
| This 46 message thread spans 2 pages: < < 46 ( 1  ) |