skyscraper or a small square in your sidebar, you’d have to start with one type of ad unit, collect results for at least a week to make sure that they’re representative, replace that unit with the second type, follow those results for a week and compare.
The result will be that the two ads are rotated randomly so that each will appear half the time. As long as those two ad units are similar in every respect but one and each has a unique channel name, you’ll be able to see exactly which type of ad unit is earning more after about a week or so.
This is an extremely useful exception to AdSense’s rules about changing its ad code, but I wouldn’t recommend that you do it across your entire site. It’s always best to do your testing on a separate page or group of pages and then make the changes across the site once you’ve got the information you want.
When you create your ad code, you’ll be asked to give that code a name. Make sure that name tells you exactly where the ad will be placed and its format.
You’re going to be looking at a lot of numbers. But two are more important than all the others.
The first is your clickthrough rate, the percentage of impressions that convert into clicks.
buying mode than users looking for something to read.
That doesn’t mean though that you should rush out and delete all of your informational sites and create product sites instead. The cost per click can
On the other hand, regular visitors are more likely to buy your affiliate products and send you more of their friends, so you really should be looking for both kinds of visitors.
There is another figure though, that’s even more important than CTR.
Once you get to grips with the numbers that you see on the stats pages and your logs, you might notice something interesting. You might see for example, that you’re getting 5,000 ad clicks on a page each month and that page is generating $1,500.
Divide $1500 into 5,000 clicks and you’ll realize that each click for that type of content is bringing you 30 cents.
The first problem with arbitrage is that you can never get a 100% CTR. Not every 5 cent click you buy is going to give you 30 cents back — and every impression that doesn’t result in an ad click is going to eat into your profits.
With these kinds of figures (and obviously, yours are going to be different), you’d need a 16% CTR to break even. (If every ad click costs 5 cents and gives you 30 cents, you can afford to lose five out of every six clicks, or 16%, before you’re in the red).
And spotting the opportunities will take a lot of effort and some headsplitting math.
Arbitrage works by exploiting gaps in bid prices for advertising space for select keywords. Advertisers looking to promote their sites that sell “grass fertilizer,” for example, might have entered the following bids:
Pay 76 cents for a click on an AdWords unit then and that user could be earning you anything from 80 cents ($1.55 - $0.75) to $1.62 ($2.37 - $0.75), minus the losses from users who don’t click the ads, of course.
Google though doesn’t reveal all of the bid prices it receives. Throw a
keyword into the AdWords traffic estimator and you can see the average estimated CPC. (For “grass fertilizer” the range was actually a much narrower $1.07 to $1.47).