 |
|
|
News From Audio Graphics:
|
A Radio Industry Step to Testing |
More than enough has been written about the radio industry's latest attempt at promoting itself. I've not read a positive review yet and don't see any reason to expect one. Kurt Hanson and Mark Ramsey voice reasons why this example of radio's creativity is destined to fail.
Rather than talk about the negatives, let's use this time to demonstrate how testing ads could have saved egg from the faces of the radio executives who gave the nod to the Radio Heard Here campaign's logo. Let's use this logo to show how A/B testing is done in an internet environment, and how easy it is to set up.
Before the demo begins, here are the caveats:
1) I am not a graphics artist. I still use :-) in my emails. I did, however, for the sake of testing, create an alternative ad to what was officially introduced. It will probably pull less response, but this exercise is about setting up the test and not the results.
|
Radio Industry Produced Logo Ad |
|
My attempt at artwork... hold the comments! |
|
2) Unless I get a Cease and Desist, this "test" will run for two weeks. That should be more than enough time to gather at least a few thousand impressions and give statistically sound samples for each version. Each ad is the same size, and links to the official "Radio Heard Here!" web site.
3) The testing ground is Audio Graphics' own RadioRow.com. It sits high in search engine returns, gets well over 100,000 page views each month, and I own the server (so I can install/remove the test code without going through a third party). AG has a number of web sites like RadioRow, used for testing new online concepts.
How to Conduct a Simple A/B Test
What follows is the process of implementing a very simple online banner ad test. Soon you will need to begin offering similar banner ad tests on your radio station web site; testing is becoming a very normal process in the world of internet advertising.
There are a number of these testing programs that we could have chosen from, but Google's is best known and most used. To begin, we go to the "experiments list" page for Google's Website Optimizer and create our experiment. (Click on any image to enlarge.)
|
Start at the Website Optimizer Experiment List, and click "Create another experiment." |
|
 |
|
|
Choose from an A/B Test
or Multivariant Test
We'll choose the simple A/B Test. |
|
 |
|
|
Make sure you have the original version, the test version, and a conversion page to verify response. |
|
 |
|
Once you have confirmed the above, it's time to name your experiment and lock in the graphics and URLs.
|
|
You will probably need to turn this over to your web master, but this part is straight forward: It asks if you will install the code. |
|
 |
|
|
|
Name your test and assign the URLs for each of the two variations and your conversion page. |
|
 |
|
|
|
This is where you retrieve the JavaScript code that makes the test functional. Google will serve 50% of the visitors version "A", and the other 50% version "B." A cookie will make sure visitors see the same version each time they return to your web site. |
|
 |
|
|
|
Once you have placed the JavaScript code on each page, the program automatically validates the test. |
|
 |
|
|
|
All that's left is to start testing. |
|
 |
|
Keep in mind that the mechanics behind testing are simple. These are the two versions of "Radio Heard Here" being tested (below).
In a couple of weeks we'll show the process of retrieving results, which is only two steps. If testing for a client, it would be at this time that you would select the better performing ad and run it 100%; then, create another version to see if you can improve the results even more.
|
This is the original Radio Heard Here banner ad logo placed at RadioRow.com. (Remember: When you go there, you will be served one of these two ads. Then, due to a cookie placed on your computer, each time you revisit you will see the same ad.) |
|
 |
|
|
This is the alternative banner ad logo placed at RadioRow.com. |
|
 |
|
That's it. Eight steps and you've set up a test to determine which logo ad generates more response. This appears more complicated than it actually is, but that's only because these are new grounds you're walking. Thousands of professionals conduct tests like these every day, and tens of thousands more professionals will be implementing tests in the future as all advertising becomes more accountable.
If your radio station web site is not already involved in experimenting with testing, it's time to start.
Seeing the results in solid statistics, and how a winner and a loser are clearly defined, gives you something to offer clients that they may not get with other traditional media. The radio industry will then be able to deliver response in a way it's never done before.
To register, click "Learn Basic Internet Analytics and Metrics." Limited Seating
|
|
Comments may be published.
|
|
|
|
