Images in GDN Display Ads: Recent Findings

Like most other PPC pros, we’re continuing to test and experiment with the GDN as we come to grips with changes, such as the elimination of standard text ads and the roll out of responsive ads.

Some of these tests are planned. But some emerge organically as our clients put forth new ideas or new challenges arise.

This happened recently when one of our clients developed a new set of GDN display ads.

GDN display ads

What Images Should You Use in Display Ads?

A few months ago, one of our clients developed a new batch of display ads. Some of these ads featured a stock image drawing of a pen. And some ads had no images at all.

At first, I thought it was a mistake.

When I realized it wasn’t, I was a little surprised—especially given our past history with this account. Previously, we had tested two types of ad images to see which performed better: images of buildings versus images of people. Our tests found that images of people performed better, so that’s what we had continued with.

But past experience has taught us that PPC “rules” that were true yesterday aren’t always true today. Things can change. So one of our senior team members, Chelsea Tryon, did some testing.

Display Ads With Animated Objects vs. Photos of People

First, Chelsea compared the “people” ads with the “pen” ads. We found that the people ads had a better CTR than the pen ads (.017% vs. 0.22%).

We can theorize as to why. Maybe users found the people ads more engaging. Maybe a photo is better than a drawing. Or maybe the ad colors were more eye catching. Or maybe people could relate more to people than pens.

Our testing also found that the people ads were generating more leads. In fact, they were generating four leads for every one lead the pen ad generated.

Of course, we have to balance this statistic with the fact that higher CTR ads will have more opportunities to convert via remarketing.

To explain: Say you search online for a jacket. Later, you’re presented with remarketing ads that feature scarves from the same brand. You might decide to click—not because you want to buy a scarf but because you recognize the brand and want to return to searching for a jacket.

So these clicks may not represent precisely what you think they do.

Display Ads With No Image vs. People Image

We also decided to test the ads with no images. When we compared them with the people ads, we found that the CTRs where close but slightly worse for the people ads (.10% vs. .08%).

However, at the same time, the people ads were impressioning about twice as often as the no image ads. In other words, Google seemed to be favoring ads with images over ads with no images.

This wasn’t surprising as Google basically assumes that images will used in display ads. It’s part of Google’s best practices.

In fact, Google talks about the importance of balancing text and images in display ads:

Balancing text and images in GDN display ads

We did find that “no image” ads and “people image” ads generated the same number of leads. But again, this isn’t the most reliable indicator of performance.


So what can we learn from this experiment?

1. Use images in your GDN display ads

Google clearly favors display ads with images over those without images. Images with photos (instead of drawings) and with crisp colors might be better yet.

2. Choose images of people over images (or drawings) of objects

People can relate to people more than they can relate to objects, whether it’s a pen or a building (even a very attractive one).

3. Always test!

As we’ve learned over and over again, you can’t generalize past experience (or even current experience in a different account) to any account. You have to test!

When you test, you’ll find that your test results will often confirm your hunches. But sometimes they’ll refute them. And opportunities can lie therein.

What have been your recent experiences with display ads?