Best Practices For Choosing A Compelling Thumbnail
A/B Testing 101, going back to the basics, and unlocking even more performance in your tests.
Optimizing your Facebook posts with Naytev drives a proven engagement lift (see our case studies), and in this three part series I’ll provide the best practices clients use to achieve these results.
It can seem daunting to get started with A/B testing, but adhering to a few battle-tested practices will ensure you see great results quickly.
This first post is dedicated to images. Simply put, better images drive better performance. If your tests aren’t yielding the results you’re hoping for, improving imagery can often turn things around.
“One of the first documented uses of the phrase.”
“A picture is worth a thousand words,” and in today’s social media landscape this is especially true. Instagram, Snapchat and Facebook, are all extremely visual mediums, so you want to make sure you have a good photo that grabs your audience’s attention. According to Brain Rules author John Medina,in the age of information, vision trumps all other senses.
- We are incredible at remembering pictures. Hear a piece of information, and three days later you’ll remember 10% of it. Add a picture and you’ll remember 65%!
- Pictures beat text as well, in part because reading is so inefficient for us. Our brain sees words as lots of tiny pictures, and we have to identify certain features in the letters to be able to read them. That takes time.
That’s why imagery on Facebook and other social networks is so important. Your readers are typically scrolling so fast through their feeds that they only stop when an interesting image catches their attention. This is reason we highly recommend testing 2-4 meaningfully different, high quality sets of images in each Facebook image, link, and video post test.
What makes a great thumbnail image?
To dive into the heart of the matter, we’re going to examine Facebook testing best practices as they apply to this example article from Reuters: Downward Facing Goat: Yoga Trend Draws Flock to New Hampshire Farm.
For this example, imagine that we don’t have access to the images from the article and have to turn to alternative sources. I’ve preselected some image sets for you to choose from. (If you need ideas for finding free high quality images, check out this article)
Given the images we’ve selected, try to guess which set of images will drive more engagement:
Are you ready for the answer?
The answer is neither!
Both sets of images will generate similarly underwhelming test results. Why? The goats are different in each image. There are aesthetic differences. There are different filters applied.
So, why will the test results be so undifferentiated?
Use Meaningfully Different Images
Ultimately the two sets of images above aren’t meaningfully different. While there are some minor variations, overall they’re similar in subject matter, composition, and the story they tell. The images below, on the other hand, are meaningfully different.
This set of images will yield better test performance because each image offers a different story and conveys a different emotion. The first image shows conflict, the second implies curiosity, and the last one is an image of relaxation (or boredom; the different possible interpretations are what make testing so important).
Each alters the presentation of the article because they evoke different moods and package the story in a different way. This kind of variation produces strong test results.
Embrace Images that Give Context
Going a little deeper into the “meaningfully different image problem”, another pitfall is the “Product Shot” Problem, which is a subcategory of providing context. Whether it’s an article about the new iphone or a new political initiative announced by Trump, people default to the standard picture of the immediate thing at hand.
In the examples above, try an image of a person using iphone to take a picture of a friend instead of stock iphone product shot; if it’s Trump, don’t just show a picture of Trump’s face scowling, show a picture of upset people, riots, doomsday, etc.
Any time your picture is of a singular person or thing (gadget, building, etc.) try images that add context. Knowing that, what do you think is wrong with the following set?
Just like before, the images aren’t meaningfully different. If you’re trying to showcase a product, use images that showcase the benefits and context of use (aka, a “hero shot”).
Use High Quality Images
Another thing to keep in mind is the quality of your images. Blurry, out of focus (unless it’s artistic), and low-res images have a much lower engagement rate with audiences. Stock Photos generally have a bad rap, but we at Naytev love sites like Pexels, Pixabay, and Wikimedia Commons. They offer myriad photo options spanning a huge range of moods, subjects, and topics.
All the images used in the examples above are royalty free and taken from our favorite sites. We have two blogposts that dive into the free resources we love and use:
- Coffee Cups, Country Lanes, and Everybody’s Cat
- Why You Must Break Up With Your Lame Stock Photos Immediately.
Also, don’t reuse photos from old tests or past posts. Always start fresh for the best results.
Use Good Variation (But not too much!)
If you find a series of images that are high resolution, resonate well, and are meaningfully different….cherry pick the best ones. On the other side of not having enough image variation, using too many image variations for a single test wastes time and money without generating meaningfully better results. Through thousands of tests, Naytev has found that the ultimate “sweet spot” for testing is…
You want to find the best message amongst all your variations, but you don’t want to spread yourself too thin.
To recap, the three pillars of good image choice are:
- Use Meaningfully Different Images
- Use High Quality Images
- Use 2-4 Image Variations per Test
But what about post text and titles? How do they play into all of this?
Stay tuned for part two to find out.
Follow us on Twitter and Facebook to get the latest updates about Naytev! Feel free to reach out to us via the support tab or firstname.lastname@example.org, we’re always here for you.