by John Eberhard

In pay-per-click advertising, one of the most useful features of Google AdWords is called the A-B test. This is a capability that you can use to set up two different ads, running simultaneously, for the same set of keywords. This allows you to test new ideas and watch the statistics for them.

Here’s how this works. Let’s say you have started up your Google AdWords account, selected a list of 30 keywords, written your text ad (which appears on Google when someone types in one of those keywords), and put up a landing page. Your landing page is a page on your web site for people to go to when they click on the ad, but as I have explained elsewhere, it should not be your home page.

So now from looking at the Google AdWords interface, you can see how many total people typed in one of your keywords (this is called “Impressions), how many people clicked on your ad and thus arrived at your web site, and what the clickthrough percentage is. The clickthrough percentage is the percentage of people who clicked on the ad compared to the total number of people who saw it (Impressions).

Usually your landing page will have a form on it for people to fill out, either to buy something or to request more information or request something you’re offering. Once people fill out that form they come to another page which is usually called a “thank you” page. You can put code from Google onto that thank you page, so that it will then send data back to Google AdWords for every person who fills out the form. This is called the “conversion code.”

If you set up your thank you page or pages with Google conversion code, now the Google interface will also show the number of conversions you got from each ad. This now gives you a very complete set of data to use for an A-B test. You’ll be able to see:

Clickthrough rate or percentage
Amount of money spent in a given time period
Number of conversions (people who filled out your form)
Conversion percentage
Cost per conversion

Now that you have your ad up there, you can now run another ad at the same time. Let’s say your first ad was:

Trips to the Moon Cheap
Fulfill your lifelong dream to
Visit the moon. Bring the family.

Now you want to put up another ad and test it against the first ad. Your new ad is:

Affordable Moon Excursion
Bring the whole family to the moon
for spring break. Family packages.

You can set it up so that your new ad has people land on the same landing page, or you can create a new landing page for it. It depends on whether your new ad is a lot different from your first ad. You must make sure that your ad and your landing page are saying the same thing. So if you change the message a lot for your new ad, make a new landing page. For the example above, you could use the same landing page.

OK, now you’ve got two ads running. You can set the system so it will show the ad that gets the best response more frequently, or you can set it so it shows them the same amount of time, which I think is best.

Now you should leave those ads running together for anywhere from 4-7 days. After that period of time, you should have a good idea of what is happening and which ad is working better. When you look at the statistics, you should set it so you are looking at them just for the period of time since you set up the second ad. That way you’ll be looking at statistics for both ads for the same period of time.

Now you want to compare the two ads, and look at such things as the clickthrough percentage rate, the conversion rate, and the cost per conversion. You will likely find that one of your ads is doing significantly better than the other one. If so, you may want to pause your more poorly performing ad and let the better one run and get all the traffic.

But let’s say they are both doing roughly the same, and let’s say the clickthrough rate was poor, say under 1.0%. In this case you might want to write another ad and run it now along with the other two. Sort of an “A-B-C” test so to speak. Or you can put up several ads. You can run as many as you want. But I would not generally run more than say 6 ads at once.

To give you some ideas of the possibilities of how you can test things, consider these scenarios:

Run text ad A with landing page 1
Run text ad B with landing page 1

So above you’re running two different text ads, but they both go to the same landing page.

Run text ad A with landing page 1
Run text ad A with landing page 2

Now above you’re running the same text ad, but going to two different landing pages. So in that case you are testing two landing pages against each other.

Run text ad A with landing page 1
Run text ad B with landing page 2

Now above you’d be running two different text ads, each of which goes to its own landing page.

I think this is a brilliant feature of Google AdWords. You can use this for a variety of purposes and scenarios, such as:

a.You have an ad running and it’s doing poorly in terms of clickthroughs, so you want to test a new ad against the current one.
b.You have an ad that is doing well on clickthroughs, but the landing page is getting poor conversions. So you test the same ad but with a different landing page.
c.You have an ad and landing page that are both doing well, but you feel you can improve the statistics even more. So you put up a test ad and landing page for a week or so, and compare against your original ad and landing page.

Run intelligently, A-B tests can be used to improve any Google AdWords campaign.

All this is still very true in 2024.