7 Best Practices For Split-Testing Your Ad Copy

White Shark Media9 years ago

[display-name-category]

[post_author]

Ads are one of the most important parts of your AdWords campaign. Your ads are basically the doors to your website with a Welcome Sign telling potential customers “come in and shop here.” This “Welcome Sign” isn’t perfect just yet; you need to test it out. The best way to do this is by Split-testing (A/B testing) your ad copies.

Here are 7 simple tips that can lead you to the right ad copy and make your AdWords efforts successful.

1. Test Out Slight Differences

You don’t want the ads to look completely different from one another. You could use two different headlines and have the rest of the ad copy read the same, or test different descriptions or display urls. Look at the examples below.

Example 1

Ad Copy A

Ad Copy B

Example 2

Ad Copy A

Ad Copy B

Once you have it them ready, let them run simultaneously and check which ad copy got the most clicks or conversions; whichever metric is the most important to you. I guess you can compare this to going on a big date; you want to know which accessory looks better with your outfit, or which cologne smells best, and the audience represents your friends helping you decide which one is the winner.

2. Have More Than One Set of Ads per Product / Service

Let’s say you have four ad groups leading to the same product and you want to have at least two sets of ads. When doing so, try to have two different descriptions on line one and four slightly different descriptions on line two.

Once you have a good amount of data (I would say wait 3 to 4 weeks to collect enough), you can download the information and create a Pivot table that can give you an idea of which description lines work best.

As you can see on the example above, there are two sets of description lines that work best; this example is in terms of conversions. You can add these ads to the other ad groups (same product/service) and later one test them out by considering tip number one.

3. Don’t Stop Testing

Your ad copies should always be tested to make sure they’re fresh. Always keep the same headlines and description lines that have proven to be the best options in the past and make small changes to the rest of the ads.

Even small changes can help you discover performance boosts in your ads that you wouldn’t have discovered if you had just let the ad copy be.

4. Ad Rotation Setting

You always want to give each ad copy within your ad groups about the same amount of time served, this way the data is a bit more even. In order to make this happen, make sure that under the Settings tab > Ad Delivery you have chosen the radio button ‘Rotate Indefinitely’.

I don’t recommend you choose the options to optimize for clicks or conversions, because this means Google will automatically show the best performing ad copy more frequently, which will not give the other ad copy a fair chance. Remember, you want your ads to be shown as evenly as possible.

5. How Long Should I Split Test?

When making slight changes on ads in a brand new campaign, I suggest that you split test slight changes weekly for the next four weeks. After that, you should leave the best ad copy running and then make changes to them every month, or when you see that conversions or CTR have dropped (these metrics depend on your goal).

6. Which Metrics Should Be Considered?

The main metric to consider depends on your main goal. If your main goal is to improve sales, sign ups or leads, then you should focus on Conversions; which ad copy is converting best. If you main goal is branding or visits to your site, then you should focus on Click through Rate (CTR). The higher the CTR means you are capturing a larger audience.

7. Which Ad Becomes the King?

You should always keep running the best performing ad. Your best performing ad is the one that has brought in the most conversions or has the best CTR, it all comes down to what your goal is. By keeping these tips in mind you’ll be on the AdWords yellow brick road to success.

Downloadable content