-
What’s Your Method of A/B Testing Ad Copy on Google Ads?
Hey guys. For A/B testing ads, I’ve read up on a lot and decided to rotate 4 ads on (optimize), and let Google Ads determine the ads with the best CTR. I would then delete the lower-performing ads (Bottom 2-3) and rotate in fresh ads with copy to try and beat my best-performing.
​
Do any of you have any alternative / better methods of determining the best ad-copy? And if you’re already doing something similar; how many impressions should you let each ad gather before swapping in new ads with certainty?
-
9 Replies
-
I usually do pretty much what you describe but recently I’ve been testing out Ad Variations under Drafts & Experiments and it seems to be working pretty well and it’s much less time consuming.
-
Well any test I’d run for at least 2-3 weeks before I can determine if the outcome is good or bad.
Secondly, I use RSAs that automatically creates the best ad copies, so I don’t have to worry about creating different ads. Best is to have 2 ETAs to see if your ad copies are good and 1 RSAa to test all other variations. -
Here’s a script that will do the heavy lifting for you in regards to A/B testing your ads (just follow the simple instructions for setup): [BrainLabs | Rotating Ad Copy Script](https://www.brainlabsdigital.com/rotating-ad-copy/)
Also make sure the “[auto-apply ad suggestions](https://support.google.com/google-ads/answer/7498247?hl=en)” account setting is disabled and set ad rotation settings to “[rotate indefinitely](https://support.google.com/google-ads/answer/112876?hl=en).”
-
Unrelated but whenever you add let’s say 2-3 new ads does the system let those new ads have a chance since the winning ad has already been selected?
-
To add a bit here – I’d consider optimizing based on Conv. Rate or CPA rather than CTR. At the end of the day clicks don’t equal sales, and if you’re optimizing only towards clicks and not sales, you’re going to have a bad time!
Someone from our team recently wrote a piece on 31 tips for Google ad copy writing / testing if you’d like to give it a read!
(Shameless plug: https://www.esperinbound.com/blog/google-ad-testing-copywriting-31-tips-for-winning-ad-copy)
-
Usually have a similar approach to most of you have prob highlighted.
Create at least 1 RSA and at least 3 ETAs for each ad group. Then create new ones over time, based on the top performing variations.
Use experiments if client wants to do a test, though not very often anymore.
-
How can you do this, but optimizing for cost-per-conversion or ROAS, and how do you do it at scale with say several hundred or thousand ad groups?
-
We use Drafts & Experiments and pre-calculate the amount of impressions needed to achieve significance on our key metric (usually conversions per impression).
-
I rotate the ads evenly, so they each have a fair shake. I don’t personally want the Google Algo’s messing with that. I can easily see which ad has the better CPC and pause the poor performing ads accordingly. I don’t need Google’s Algo for that.
Log in to reply.