« Creeping Credibility for the Internet | Main | More On Click Fraud »
April 11, 2005
Click Fraud Was Inevitable
There has been a lot of flurry lately about click fraud, but it was inevitable because click-through was never a good metric for measuring the effectiveness of online advertising.
The Wall Street Journal had a front-page story on Wednesday, April 6, by Kevin Delaney, titled,” In ‘Click Fraud,’ Web Outfits Have a Costly Problem” and a sub-head that read, “Marketers Worry About Bills Inflated by People Gaming The Search-Ad System.” Delaney’s story goes on to explain what click fraud is (unscrupulous companies send out spiders to run up unproductive clicks on competitors’ search ads) and how it works. The reporter gives examples of refunds one advertiser got from Yahoo and Google because of “unusual clicks” and “invalid click activity.”
Delaney writes, “Click fraud is the latest problem plaguing the Internet, alongside spam, identity theft and online-auction fraud. Some believe about 20% if clicks are from people not necessarily interested in the product advertised, and therefore in the industry’s view, fraudulent.” But click fraud was inevitable because pricing Internet advertising based on click-through is and always has been wrongheaded—it not only encourages fraud but also undervalues Internet advertising.
There are several methods of pricing Internet advertising, as author and AOL sales executive Vince Thompson details in his chapter on Interactive selling in my textbook, Media Selling: Cost-per-thousand impressions (CPM), cost-per-click (CPC), cost-per-acquisition, cost-per-registration (CPR), cost-per-trial (CPT). All but CPM are performance–based pricing models, which advertisers prefer because they transfer the risk of advertising from advertisers to Web publishers. Performance-based pricing also fails to give any value to either the branding or awareness value of an Interactive ad, whether it’s a search link or a 160 X 600 pixel tower ad (standard Web units are defined and labeled by the Interactive Advertising Bureau).
By pushing the risk off to publishers and assigning no branding value to an ad impression, advertisers attempt to keep Interactive ad prices down, which is understandable, but it is also penny wise and pound foolish. Clicks don’t mean sales and all Internet ads, even one-line search ads on Google or Yahoo, have some branding value, which is why publishers such as AOL, MSN, and Yahoo (except for search ads) prefer CPM or other non-performance-based pricing models such as sponsorships, revenue sharing, or monthly placement fees.
When advertisers buy advertising other than search ads on a CPC model, they push risk off to publishers and typically don’t pay attention to creative execution, don’t change the creative enough, and simply don’t pay attention to their schedules, figuring that if no one clicks it’s the publisher’s problem because the publishers won’t get paid if no one clicks. When the CPC model is applied to search ads, advertisers often do not experiment with the small amount of copy in a search ad and, thus, miss an opportunity to increase clicks with better copy. Also, as we have seen, a CPC model leads to gaming the system because unscrupulous competitors can, and do, use spiders and other methods to substantially increase the clicks on competitors’ search ads, thus driving up their cost of advertising.
What pricing models are better than cost-per-click? Cost-per-acquisition models are better but are hell to administer and also incentivize cheating. Let’s say someone clicks on a search ad and buys something (and airline ticket, for example), then the airline has to keep track of all the acquisitions and pay the publisher on a monthly basis. But what if the buyer cancels the ticket or upgrades, the airline has to keep track of it all and there is a tendency (an incentive) to underreport transactions. A cost-per-impressions (CPM) model is best in my view because a CPM pricing model puts the risk of advertising on the advertiser, where it belongs, and, implicitly, monitizes the value of branding. In other words, if an Interactive ad runs, it has value regardless if anyone clicks on it or not.
I believe the current click-fraud problems Yahoo, Google, and their search advertisers are having will lead them to go to a CPM pricing model within a year. Because Yahoo has been more innovative on the sales side of the business and is run by a more business- savvy CEO, Terry Semel, than Google, I believe Yahoo will be the first to change its search pricing model to CPM, which will have a positive impact on its profits.
Posted by Charles Warner at April 11, 2005 02:06 AM
Trackback Pings
TrackBack URL for this entry:
http://www.mediacurmudgeon.com/movabletype/MT/mt-tb.cgi/182
Comments
Charles Warner at April 15, 2005 02:54 PM writes:
Thanks for your intelligent comment, gregbo. I tend to agree that some form of flat or sponsorship rate not based on either impressions or clicks would be better for publishers, but I think advertisers would baulk on non-performance measurements. Perhaps a rate based on unique visitors would be acceptable to advetisers--that metric would be similar to magazines, which have a rate base based on audited circulation.
gregbo at April 13, 2005 11:37 PM writes:
In general, I agree with you, and have been saying this for at least
eight years. However, I don't see how CPM is any better than CPC,
because automated bots can still be written that generate lots of
impressions. My general feeling is that the SEs should just charge a
fixed fee per unit time for each ad. They do not have to guarantee
any # of impressions, nor do they have to limit the advertiser's
ability to run their own searches to check for placement, etc. If the
advertiser doesn't see a substantial increase in sales, they can
always discontinue the ad. Also, this will reduce the costs on the SE
of implementing fraud detection. The only question is how much money
should the advertiser pay per ad. They can still bid on keywords, if
they wish, although I imagine some advertisers would rather that their
ads be placed regardless of whether the ad is relevant to the search.
The SEs could allow this type of ad to run if it does not affect their
users' experience, and probably make more money, but if their users
wish only to see relevant results on the page, they'll just run
relevant ads. Still, no one is hurt.
federico riva at April 13, 2005 11:31 AM writes:
i totally agree with you. BTW, ppa models also suffer a lot for the 'cookie crisis'. Cookie are a veri inefficent way to track orders and leads. Only electronic coupon could substitute the cookie use or the 'less disabilited' SOs (Macromedia technology). Unfortunately, i guess that the ppc se will not re-switch to a ppi model, because they paradoxically gain much more with ppc. another short consideration; i think click fraud is not so 'dark' as the lack of transapency of the bids. With Google you cannot know how much do you oay more than other cusomers (and less too) and on overture and other ppc se you cannot know if the shown prices are real or not. so you can oay 1 cent more than one who is really paying 50 cents less than you.
Post a comment
Thanks for signing in, . Now you can comment. (sign out)
(If you haven't left a comment here before, you may need to be approved by the site owner before your comment will appear. Until then, it won't appear on the entry. Thanks for waiting.)Printer-Friendly