Tests perform best when you try a minimum of 2 meaningfully different images, though 3 is ideal.

Be sure to use high-quality images and images that are meaningfully different. Images with products in slightly different locations or varying filters don't have a strong effect on test results. Low-resolution images and emotionally neutral images (e.g. a picture of a standard office building) can also undermine engagement.

Take a look at these meaningfully different images regarding our test content "Goat Yoga" below:

TIP: Short on images? No problem. Read more in our blog posts:3 Fast Ways to Find the Free Stock Image You Actually Need and Flirting with CC0 Images.

Much like image rule, tests perform best when you use a minimum of 2 meaningfully different sets of text, though 3 is ideal. 

Try alternate curiosity gaps, story angles, calls to action, etc. Swapping a couple words, or reordering a sentence typically won't move the needle.

For Link Post tests, be sure to try multiple variations of both the Post Text and Headline; both greatly influence performance. We don't recommend testing different Descriptions - it rarely impacts performance, and commonly isn't seen by readers anyways (Facebook truncates it out of view on mobile, you'll see the warning pop up in the GIF). Feel free to keep this field blank, or stick with your default Open Graph description.

If you're running a video test, you have the option of testing more than 1 version of a video, in addition to images and text.

For example, you could include both a 30 second and 90 second version of the same video, or include both a landscape and portrait version of a video.

After you've taken these guidelines into account, you're ready to preview your different test variations!