Jason Puckett

March 6, 2015

Ad Copy Testing - 5 Best Practices to Improve ROI

Ad copy testing is absolutely vital for every growth marketer. Whether you are testing different ad copy for the same product/audience, or the same ad copy for various products/audiences, the entire testing process can become very difficult without a concrete approach.

Automating this process can make testing easier, but ad testing platforms are only useful if you have the proper fundamentals to apply. Here are five best practices that can lead to improved ROI numbers from your paid acquisition channels:

Plan Before You Act. Make Ads Hugely Different.

Know what you’re testing. You are a marketer who has ideas in your head about what will work and you need to prove or disprove these ideas. That is what testing is used for. Through your years of experience, you’ve always wondered “what would happen if I tested X,Y, or Z?” These are the ideas to start with.

Big changes will show obvious results. Start by making headlines drastically different; use numbers vs. letters, use completely different phrasing in descriptions or very different display URLs. The larger the difference in the ad copy, the more noticeable your test results will be.

Know which metrics you are going to measure, and make a hypothesis before you launch your ad test.

Create Purchase Scenarios

You know your audiences. You know your products. As a marketer, plan out tests for your top three audiences and be very specific about them. When we say “purchasing scenario” we mean a realistic situation where someone will convert. An example: “Female. 18-30 years old. Lives in the US. Has been to our site before”. That is a great scenario for a remarketing ad test. Go figure out which ad content drives her to buy… or your business’ version of her.

Use a Z-test to Calculate Statistical Significance

Find an ad in your AdWords account that converts really well for your determined “purchasing scenario”. Use this ad as the baseline in your ad test. In order to prevent AdWords from auto-optimizing your ad copy, switch your Ad Group settings to “rotate ads evenly” and use a sample size of at least 100 clicks per variation.

Technically, a Z-test only needs a sample size of 30, but we don’t recommend basing decisions from samples of that size.

Let the Data Make Decisions For You

The use of a Statistical Significance Calculator can really help your decision making. Any variation that beats your baseline and has a confidence threshold over 90% can be determined as a winner.

These winners should receive a high percentage of the ad spend for that ad group. Phasing out existing front runners should not be an immediate process. Continuously lower the share of spend for old/underperforming ads. Immediately eliminating existing front-runners can have a negative impact on overall performance because of the history these ads contain.

Continuous Testing Means Continuous Improvement

Testing is a continuous process. Once you’ve executed a successful ad copy test for one of your products or audiences, it’s time to create another. There is always room for improvement and always another “purchasing scenario” to optimize.

Good Luck. Testing is the Basis for Success.

Have a question for us? We'd love to hear from you!

comments powered by Disqus


Popular Articles

Understanding Conversion Layers and Executing Ad Experiments

August, 14th 2014 - Optimization

3 Steps for Building a Paid Acquisition A/B Testing Plan

October, 15th 2015 - Optimization

All Things Tested - 3Q Digital's Sean McEntee

August, 31st 2015 - "All Things Tested"