» PPC » Things you can AB test in Adwords – A guide

Things you can AB test in Adwords – A guide

The following two tabs change content below.

Andy MacLean

Blogger at Andy-MacLean.net
I'm Andy MacLean, blogger & SEO consultant. If you like this blog post (pretty) please leave a comment below or share it. Thanks for reading!

AB testing in Adwords i’d guess is performed by a very small minority of Adwords users yet it’s a really valuable tool that can help maximise your Adwords results.

What things can you AB test in Adwords?

You can test a wide variety of things within Adwords experiments. Here are the tests that I think can lend most value to your campaigns:

1) Ad copy – the obvious one. Run 2 or more ads together and see which has the better CTR, conversion rate and ROI.

2) Destination url. Find out which destination url produces the best conversion rates and ROI.

3) Dynamic destination urls. I sometimes work with a developer to make elements on the page change, usually the h1 (header tag/title), by using paramaters in the destination url e.g. abc.com/example-page V abc.com/example-page?Sale-Ends-This-Week. The latter changes the standard H1 on the page to ‘Sale Ends This Week’. If you access to a savvy developer you could do the same for images, pricing and whatever else you think could be changed on the page. Example of a H1 that could be tested…


4) Bids. I’ve never heard of anyone else doing this before and perhaps it’s not well known that you can do this but you can actually AB test your max cpc’s. This can help you find out things such as; how much extra traffic you can get by being in top position, what the impact is on CPA by bidding +50%, how much more impression share you can get by increasing your bids etc etc.

How to set up AB experiments in Adwords

Part 1

1) Go to your campaign settings and hit the experiment option.

2) Give the experiment a name, hint; call it what you are testing.

3) Decide on the split ratio – there might be a reason why you don’t want a 50/50 split but it’s unlikely so choose that option.

4) Enter the dates you want the experiment to run. Here I would consider how much data the A & B elements will need to make the results significant. If you want it to run the instant you are ready to make the AB changes then choose ‘no start date (i’ll start it manually’) then you’ll have the option to hit a start/stop button once you have saved the settings. NOTE: if you choose this option don’t forget you’ll need to stop the experiment manually!

5) Go and make your AB changes.

adwords experiment settings

Part 2

Ad copy – go and create an a new ad in the ad group/s where you want to test new ads. If the experiment is already running you’ll have a tick box option to have that new ad run as experiment only, tick it. If the experiment hasn’t started, go back to settings and hit start. If you selected todays date as the start date it will be running. The existing ad in the ad group – set it to the control by toggling the lab flask icon. You will now have 1 control ad and 1 experiment ad running with each getting 50% impression share. I would stick with just 2 active ads in the ad group – you’re control and experimental ad otherwise things might get messy!


control experiment



Destination urls – repeat the above but this time you want to just change the experiment ads destination url. Tip: copy and paste the original ad, then edit the destination url and toggle the lab flasks rather than rewrite the original ad.

Bids – Hit the keywords tab in the ad group/s or campaign/s you want to AB test bids. Click on the segments dropdown and hit ‘experiment.’ You can then set a bid modifier for the experimental bids.

experimental bids

Reviewing Results of AB testing in Adwords

To review the results of your AB testing experiments in Adwords you just need to look at the regular reporting tools i.e. to look at ad copy tests hit the ad tab and see the difference in stats for your control and experimental ads. To look at your bid tests go to your keywords tab and hit the ‘experiment’ segment again.