Split testing is the basic process of trying two very similar things to see which is more effective. With Internet marketing this could be two banner ads, landing pages, email scripts, or subject lines. Split testing is a good thing to try whenever you are trying to determine the most effective copy, image, or headline for… well anything. In fact in Timothy Ferris's book, The Four Hour Work Week, he explains how he even used split testing with online ads to determine the best title for his book. Here are some basic rules I like to follow with split testing my campaigns.
- Test Everything. Test PPC campaigns, banners, emails, headlines, etc
- Have a plan to measure results. Depending on the campaign you might have a built in measurement system but sometimes you have to get creative about how you will determine which is more effective
- Test only one variable at a time. If running ads on Facebook you shouldn't run two ads that have different headlines, images, and body text. Run two ads that are the same with only one variable different so that you can actually draw a conclusion about what is and isn't effective.
- After you figure out what is effective, split test again. Just because one is better than the other doesn't mean it's the best possible option.
- Don't draw conclusions from too little data. You have to run the test through a significant amount of volume before you settle on the results.
I recently started wondering about the street beggars I pass downtown. I want to do some split testing with their business to determine the best signage, locations, accessories, facial expressions, etc. There are best practices for everything and I find myself wanting to split test everything in life. Any other ideas of what would be worth split testing?