Google's AdWords is a great way to advertise on a budget. You write an ad and pick keywords that, when searched for on Google, will bring it up. Depending on the competition for your choice of keywords, you can advertise for as little as 5 cents per click. And you don't pay a thing for impressions that aren't clicked through.

Goggle keeps track of how many times each of your ads is displayed, and how many clicks each gets. If you have multiple ads in a group, they'll display the ones with the best click-thru rates more often than the others. This helps you maximize clicks, and by viewing the statistics, shows you which ads are the most effective.

The most effective at generating clicks, that is. I've been doing a little analysis of the performance of my Google ads, and something I knew intuitively before has become crystal clear: maximizing click-thru benefits Google. Advertisers need to do a little extra tracking to maximize profits. A few factors need to be balanced against each other to do this.

First, why is click-thru maximization not always best for the advertiser? If you were paying for impressions, click-thru would be more important, but when you're paying for clicks, you don't want people clicking unless they're going to buy.

If an advertisement has high click-thru but low conversion, then it is probably not communicating your message clearly. People are seeing the ad, and it's doing a great job of capturing their interest, but when they get to your site, they're not finding what they thought they were going to find. An ad that communicates more clearly may not have as a high a click-thru rate, it will probably have a higher conversion rate.

Of course, some of the people who click through your ad and don't find what they expected will stick around and buy something else. So click-thru is where it's at after all, right? Not necessarily. A few examples will illustrate:

Example 1
1000 impressions
100 clicks (10% click-thru)
5 sales (5% conversion on click-thrus)
25 cents per click for a total of $25
$5 per sale for a total of $25
Zero profit

Reader Comment:
Antone Roundy said:
Another example to show how NOT to decide which ad is the most profitable. Let's say you're running two ads, with the following stats: Cost per click: 10 cents Income per purchase: $5.00 Ad 1: Impressions: 900 Clicks: 90 Cost: $9.00 Purchases: 9 In...
(join the conversation below)

Example 2
1000 impressions
50 clicks (5% click-thru)
3 sales (6% conversion on click-thrus)
25 cents per click for a total of $12.50
$5 per sale for a total of $15
$2.50 profit

Of course, if you make more than $5 per sale, your ad will make a profit, and if your profit per sale is high enough, you might make more profit than on the ad with a higher conversion rate. The point is that your profits can be higher even if your click-thru rate is lower. So you need to do the math and figure out which of your ads is generating the most profits.

To track that, have each ad point to a different URL. That can either be a completely different page, or you could just add something to the end of each URL to identify the ad. I add "?src=g1", "?src=g2", etc. to my ads, for example, "http://www.geckotribe.com/rss/jawfish/?src=g1" ("g" stands for "Google"--I use a different letter for each source). Then, when someone arrives at my site from Google, I keep track of where they came from, and if they sign up or make a purchase, I know which ad was responsible.

A recent analysis of 5 Google ads showed that all were profitable, but the profit margin for the best was 24 times as high as the worst. I don't think that the difference is entirely due to differences in the ads, because one had been running longer than the other, and my conversion rate has changed over time. But this information gave me valuable information to use to adjust my advertising mix and see if I can increase my profits.

I deleted the three least profitable ads and now have just the two best running. Maybe I'll find out that the change in conversion rate was a result of different ads, and my profits will jump. If not, I'll write a few more ads, track the results, and try again.