Eye Tracking studies can KILL conversions if you focus on the wrong things for your landing page optimization tests!

Posted by 5 Comments

Eye Tracking studies can KILL conversions if you focus on the wrong things for your landing page optimization tests!

Conventional wisdom states: You should use images that guide a user to the actionable elements (or interest/desire elements) on the page. That’s what we’ve been told and that is what most of us have believed up until now. This idea stemmed from eye tracking studies where an image of a person looking or pointing at your call-to-action guides the user to look in the same direction, and it works! Doing this is great if you want your users to look at your call-to-action rather than click on it.

Eye Tracking studies can KILL conversions if you focus on the wrong things for your landing page optimization tests!

Eyes Can Guide Your Eyes

Just look at this cool image from Human Tracking. Clearly the eyes glanced to the left will guide the viewers eyes to the left.

Guiding potential users to your call-to-action using page elements or images is common practice in conversion optimization. But we hypothesized that, while guiding a user to a button can increase conversions, we had a belief that conveying emotion is more important than simply guiding the eyes of the viewer.

Now I know some you are saying to yourself “Duh!” But look around… Most sites focus on guiding the eyes, not conveying emotion. Since we are heavy on “copy” in our company, clearly emotion is something we yearn to elicit in people who visit our sites.

Before we go on, I want to give a BIG shout-out to our massive-traffic clients who let us do crazy tests to learn from and to grow their conversions. Without them (and the massive amount of traffic) we couldn’t do these types of tests. And just FYI, a little disclaimer: No landing pages were harmed in the making of this blog post. In fact, we were able to chalk-up another W in the win column for our incredible clients. (Whew!)

Emotion vs. Eyetracking


To find out, we created a test where we replaced nothing but a single image of a woman on a current landing page, the rest of the page was kept exactly the same for the sake of control and valid test results (something a great post from 37 Signals can’t say about their latest conversion test). (Hint: They changed photos AND copy. Although a great design test.) And to clear up on statistical issues, our test was conducted with over 150,000 unique visitors, and 95%+ confidence levels on the versions that hit over 4% difference.

We tested 10 different images, all which were of the same model wearing the same color posing with different emotions and looking or pointing in different directions. Don’t get me wrong, eye tracking and creating a well designed eye-path is GREAT, but too often that’s all designers focus on.  While the images of the woman looking and pointing at the call-to-action on the page gave a slight conversion increase, they were not the highest converting images. Surprised? You should be!

It turns out that our hypothesis was correct. The model’s emotions were a larger factor in the conversion increase than where she was guiding the eyes of the user. We believe this is because the model’s emotion elicits a similar emotion in the user and in doing so, we can help the user into the emotional state that best fits our service or product and, in turn, makes the user more likely to make take an action.

For our test, we used this the following image as the baseline to compare all other images against. We chose this image specifically for being the “perfect” eye-tracking study image. The woman is smiling, looking and pointing towards the call-to-action on the page. All test images are being compared to this baseline.

Eye Tracking studies can KILL conversions if you focus on the wrong things for your landing page optimization tests!

Of the 10 images that we tested, six of them were of the woman smiling but not excited. Five of those were looking or pointing towards the call to action item on the page. These should be the highest increase in conversion rates, right?

Hover over the images below to see their results compared to the baseline image.

Eye Tracking studies can KILL conversions if you focus on the wrong things for your landing page optimization tests!

The images in red show a negative conversion rate change, the yellow images show a small (5%).

Also during this test, we found four outlining categories that the test images fell into: excited-forward, excited-pointing, plain-forward, plain-pointing. (The plain pointing category was the baseline so all conversion rate increases or decreases are in comparison to this.)

If you hover over the following chart you will find that while the plain-pointing is better than the plain-forward, both the excited-forward and excited-pointing are higher than both plain categories. The most surprising find is that, in our test, the excited-forward performed better than the plain-pointing category.

Eye Tracking studies can KILL conversions if you focus on the wrong things for your landing page optimization tests!

Note: CR #’s changed because the baseline is now a group of images rather than the single image first shown above.

Final note and thoughts. We’re big fans of eye-tracking studies and seeing how people are visually responding to our designs, and we also recommend you become a fan of understanding how people view your site. (BTW, we like Gazehawk). The biggest take away for us, and serves more as a reminder: conveying and eliciting emotion is key to conversions. And if you can guide the eyes, as well as guide the emotions, double bonus points, life up!

Another interesting note from the data that is not shown above, is that in tracking the “value” of the conversion through the entire profitability funnel for our client, we noticed that in addition to the conversion rates being higher, the “value per unique” for Image #6 & 7 above had 10% and 12% increase over the baseline on the $/unique, and Image #8 was a -8% (Negative 8%) below the baseline on value. That’s a pretty big difference in $/unique. Just confirms the emotion theory.

What are your thoughts? What does this data tell you? What are your takeaways?

Would you like to have one of our conversion rate optimization experts review your site? Just click here to contact us today and learn how we can drastically increase your website conversion rate for no money up front, and paid only on results!

5 Responses to Eye Tracking studies can KILL conversions if you focus on the wrong things for your landing page optimization tests!

  1. Todd Barrs says:

    Excellent research & testing and thanks for sharing the results! I now have my data driven reference to point clients to when I can’t convince them that emotion is critical to include in copy and images.

    One thought I had while reading was using images & emotion to direct eyes to the headline vs. the CTA since a headline arguably invokes more emotion than a CTA (it’s a CTA — it’s the “reaction action” to emotion invoked by images and headlines). The two elements could be one in the same, in which case it’s a moot point, but still worth noting.

    Again well done. One of the best, most thorough and well-backed articles that I’ve read this year. Kudos!

  2. Dustin says:

    This study was a genius play. As a marketing major in college, I’m intrigued by the data that you collected as it shows how consumers react to particular advertisements online. Never really knew about the concept of GazeHawk.com but am definitely considering giving it a try.

    I was reading a book called “Buyology” – if you haven’t read it, the focus is on how consumers brain activity triggers consumers reactions. It’s what the book has dubbed “neuromarketing.”

    It’s official to say that marketing isn’t a game, or challenge, of targeting based on demographic, geographic location, or geodemographics anymore. It’s actually going INSIDE the minds of consumers.

    Great write-up none-the-less.

Leave a Reply

What is 4 + 2 ?
Please leave these two fields as-is:
IMPORTANT! To be able to proceed, you need to solve the following simple math (so we know that you are a human) :-)