Split Testing Image Ads

HTML clipboardJust as important as your landing page, your image ads can make or break your ad campaign. When using Google Adwords Images and advertising on social networks like Facebook and MySpace, your campaign cpc is based on your CTR. If MySpace or another network is seeing that users are clicking your ad at a much higher rate than others, you will be given that much more exposure, while having to pay less per click. It all backs out to a CPM model for these networks.

I wrote a post a couple months ago about split testing with facebook ads, and I wanted to update you with a few more test numbers and how you can improve your overall campaigns.

Here’s another campaign I ran a few months ago that was targeted towards users that had an interest in “soccer“. The text ad copy for all of the campaigns were the same, but I setup a few different ads for each campaign. The ads that were pulling the better CTR rates, was strictly because of the different images being used. Which would have drawn your interest the most?

.21 CTR
.14 CPC
.20 CTR
.15 CPC
.09 CTR
.20 CPC
.09 CTR
.21 CPC

Sometimes the images that you think will perform best, don’t… and this is why you should continually mass test and see how different images perform with your campaigns.

Another tip is to setup campaigns for specific interests. For example, if you have an offer that is for getting people subscribed to “free tv magazines“. Setup a separate campaign for people interested in “Grey’s Anatomy”, then use a Grey’s Anatomy graphic… instead of just showing the picture of the magazine or a TV. You will improve your overall numbers that much better with tighter niche groups and relevant images.

For people to actually go into their account and make an updated of what they are interested in, or a fan of… you know they actually are. Through Facebook you can target all of this so easily. People obviously like to click on what they are interested in and believe in. During the days before the election, I ran a few campaigns targeted towards people interested in Barack Obama and John McCain. Through these small initial tests, I was seeing some nice CTR rates, all the way up to .75 CTR in some groups.

It was only a few thousand clicks, but it was also only accomplished over the course of a few days before the election. The point is, you can make good money by targeting people’s interests, as they will more often click your ad, then a cross network random ad.

Keep adding new images to your rotation of ads, weed out the ones that under perform and try to find more relevant images to the ones pulling in your best ctr.

Similar Posts


  1. Awesome info. To answer your question, I definitely would be more likely to click on the far-left picture. It just looks more… interesting.

    1. That was my first thought too. The far left picture is the catchy image. Such images can increase clicks and stay time. Remember shoemoney and the blondes??

  2. nice post as usual, i'm thinking of using facebook advertising, you get anything if I sign up under you?

  3. Loved your idea on specific interest to draw segment, simple but profound. Also, am now sold on Facebook. Going to give it a try.

  4. That is fascinating! Excellent article. This is the best and only article I have seen written on image split testing. Thanks for the great education!

  5. Thanks for the great post – it's insights like these that help all your avid readers get closer and closer to profitable success on the internets.

    I subscribe to a butt-load of RSS Feeds, but yours is one of the very few that get an immediate click when I see it's been updated… while simultaneously clamping down on my pubococcygeus muscle so I don't accidentally wet my pants with glee. 😉

    Future topic suggestion (if you're drawing a blank): Your opinion on CPM Facebook ads vs. CPC advertising there.

    Regards and thanks again.

  6. Great post as usual Zack. It's a trip to how much your CTR can go up or down by only changing the pic. But hey it pays off in the end. Nice post.

  7. This is a great post Zac , but the picture seems incomplete without the Sales figure … an ad with less number of clicks could have been the one that actaully drives sales, an ad with a high CTR does not mean more sales …any thoughts ?

  8. Zac, great case study.

    I'm interested in conversion numbers rather than CTR. Got any insights on how those compare between different ads?

    1. Good point. If a very good image leads to a average landing page, the story remains cliche.. You need good landing pages, which can be propelled by good images..

  9. I couldn't have said it better myself Zac. We must always test and analyze our campaigns to find what gives us our best ROI. Don't you just love the way Facebook lets you target someone based on a certain interest. It really couldn't get much easier.

  10. Zack i'm very interested in conversion numbers. I still think text ads convert better.

    Any ideas on how video ads convert?

    This would be also extremely interesting!

  11. I just love your Blog because you always show us what others hide (secret techniques) though they are simple but they need common sense 🙂

  12. I'd be tempted to click on the very left one. Because it looks like there's some story, and there's interaction between people you can kinda relate to. I'd be clicking to check out the action, or find out if it actually tells you what's going on there in the picture.

Comments are closed.