Forum Moderators: martinibuster

Message Too Old, No Replies

More Hits = Less Pay per hit

         

dvduval

10:37 pm on Oct 10, 2004 (gmt 0)

WebmasterWorld Senior Member 10+ Year Member



For the longest time, I was only running ads at the bottom of a high traffic site (close to 1 million page views per month). Over the last two months, I have been testing running ads at the top. I'll run the ads at the top for 2 days, and then at the bottom for two days. Overall, I'm getting more hits, but the payout per hit seems to decrease when the number of hits increases. After 2 days on and 2 days off for a period of time, I believe I have enough data to be personally convinced there are some algorithms that are being used, and that I am working with a loaded deck of cards. I am going to continue testing, but if this premise turns out to be true, it is certainly going to further taint my image of the once highly respected Gooooooogle.

AZEvil

7:40 am on Oct 12, 2004 (gmt 0)

10+ Year Member



FromRocky,

Have you ever looked at how listings are ranked in the AdWords system (which is where the AdSense ads come from)? My statement is true and I have actually seen it happen on my own AdWords ads being displayed on content sites. When the CTR drops dramatically on an ad, the cost per click goes up to hold that same ranking. I have been testing the theory of higher CTR causing lower cost per click in AdSense and it does happen. The same ads are clicked, placed in the same location, and the payout per click drops dramatically when the CTR goes up (in my testing, I have been able to vary my CTR by nearly 8%). I have hundreds of daily clicks and make several hundred dollars per day from these clicks. If my CTR is 10% or above, the average payout per click can be as low as $0.50. Since I have found the "sweet spot" in the CTR for the ads I display, I have averaged around $1.80 per click. I also tested multiple ad units on a site and saw that the CTR dropped dramatically and the payout per click dropped as well. I watched the ads that were being clicked on and noticed that ads on the bottom of the page (the third ad unit) had a higher CTR, which meant that they were not paying as much. I have done away with the multiple units on almost all of my pages. I feel it's better to keep a user on my site than to lose them to a $.05 click.

Another consideration that people have to make is that once an advertiser's budget is up for the day, their ad stops displaying. If you have new competition coming into the market for displaying those ads, you will deplete the higher paying ads faster and show lower paying ads. There are so many variables in the system that any one (or combination) of then could create drops in revenue, payout per click, etc. I see no reason not to trust that Google is treating advertisers and publishers fairly. What motivation do they have to upset either side of the system? If publishers leave for better payouts, advertisers will not be able to get as many clicks on their ads. If advertisers have to pay so much per click that they have no ROI on their ads, they will pull the ads out of the system.

AZEvil

7:53 am on Oct 12, 2004 (gmt 0)

10+ Year Member



Oh...by the way...here's Google's wording directly from the AdWords site:

Pay less for more results: Keep in mind that because of our AdWords Discounter and dynamic ranking system based on CTR and CPC, your actual cost-per-click (what you actually pay per click) is often less than your maximum CPC (what you are willing to spend per click) and often decreases when your ad starts to perform better.

walrus

5:05 pm on Oct 12, 2004 (gmt 0)

10+ Year Member



Great post AZEVIL!
That explains a lot.

Macro

5:10 pm on Oct 12, 2004 (gmt 0)

WebmasterWorld Senior Member 10+ Year Member



AZEvil, I've been trying to find the flaw in your logic but I can't. I realise now that I've been trying to find the flaw because I don't have the time/can't be bothered to go through the trouble and research you did. ;)

But if this does indeed work all I can say is "WELL SPOTTED"!

This 34 message thread spans 2 pages: 34