Podcast: Play in new window | Download | Embed
Subscribe: Apple Podcasts | Google Podcasts | Spotify | Amazon Music | Stitcher | Podcast Index | TuneIn | RSS | More
In author groups on Facebook, I regularly see authors asking one another to help them choose between two book cover options.
Unfortunately, this is not a good way to learn which book cover will resonate best with your readers.
In this article, I will teach you a scientific method that will help you determine which title, cover, blurb, or tagline will be best. It’s a technique so advanced that most publishers don’t use it. But it’s also so powerful that you should learn to do it yourself.
Split testing is a technique that has been around for decades and has proven its reliability over time.
Novel Marketing patron Mary Hamilton asked about it:
On this week’s podcast, [Thomas] talked about A/B testing. Could you do an episode on that? I get the idea generally, but I don’t quite understand how it’s set up.Novel Marketing patron, Mary Hamilton
What is a split test?
Split testing was first developed in the 1950s for direct mail marketing. A business would create two versions of a sales letter. The company would mail Version A to a percentage of the people on their mailing list with a 1-800 number recipients could call to order the advertised product.
The company would also mail Version B to a different group of recipients who received a different 1-800 number to place their order.
After some time, the company would see which phone number received more calls, and then they would choose the sales letter that prompted more phone calls and mail that version to the rest of their mailing list.
Instead of asking fellow businesses which letter was better, they essentially asked their customers.
Today, it’s easier than ever to implement this method because of Facebook ads. Authors can use science and data to decide which title, cover, blurb, or tagline their readers will like best.
It’s important to know that a split test is not a survey. When you ask your Facebook friends or fellow authors about your cover, that is a survey.
Surveys Are Less Effective Than Split Tests
Your Facebook friends pay too much attention to your book cover because your opinion influences them. They want to make sure they give you good advice. They may be influenced by how many votes each cover has received so far.
That’s not how it works in the real world. When your book is on Amazon in a list of a dozen other book covers, you want yours to stand out.
There is a difference between what people genuinely like and what they click on.
You want the cover or the title that makes a reader click. Only scientific testing on tens of thousands of people can help you make that decision. And you can test relatively inexpensively.
When you have a Facebook Ad with a book cover or title that generates lots of clicks, that data is helpful because it simulates the online buying environment. Facebook is a noisy place with lots of visual distractions, just like Amazon.
What can you test with the split test method?
If you have two or three book title ideas, you can test to see which gets the most clicks. Tim Ferriss, the author of The Four-Hour Work Week, developed his title with split testing. It became a runaway New York Times bestseller and sold millions of copies. He got the title right by testing using Facebook ads.
If your designer sends you three covers you like, but you’re not sure which is best, you can easily use Facebook ads to test which is best.
You can’t actually test a paragraph because there isn’t enough room for that much text, but you can test a sentence like your hook or tagline. Blurbs are the hardest to test because of the limited space you’re given in the ad.
Titles and covers are more important and easier to test on Facebook than blurbs.
Why is testing a good idea?
You are just one person, and if you like your cover, you absolutely know one person who likes your cover and would buy it. But you want to sell more than one book.
Even if you ask friends and family, the sample size is just too small to glean any reliable data. You need to test your covers on at least 100 people to get accurate data. To minimize the margin of error, you need to test on at least 400 people.
It’s hard to read the label when you are standing inside the bottle. As the author, you’re inside the bottle of your own book, and you don’t have the perspective you need. You’re too close to it.
Don’t leave these decisions to chance. Your title and cover can make your book succeed or fail, and if people aren’t excited about your title and cover, your book is dead-on-arrival.
The Big Picture of How Split Testing Works
In a split test, you create two versions of the same ad targeting the same group of people. Everything about these two ads is identical except for the one element you’re testing.
If you’re testing the cover image, you keep the title the same in both ads and only change the image. If you’re testing the title, keep the image the same and use different titles on each ad. Run both ads at the same time. You want to target an audience of at least 100,000 people who are the kind of people who would read your book.
Facebook will run your ads simultaneously, and that same group of people will see both of your ads. Over time, your Facebook ads dashboard will show your stats. For example:
“Ad A” got 10,000 views and 100 clicks.
“Add B” got 10,000 views and 200 clicks.
That scientific data tells you that “Ad B” is the more effective ad. Use the image or the title, whichever you were testing, from “Ad B” because twice as many people clicked “Ad B.”
Expect to pay around $50 per test, so enough people see your ad. Valid results require a large test sample.
Two Principles for Running A/B Split Tests
Test One Thing at a Time
It’s important to test only one element at a time. Multi variant testing is possible, but only if you have a degree in statistics and love very complicated math.
Test the title first to get the better title. Then test the images and find the better cover.
Test on the Right People
If your book is for 50-year-old women, make sure your ad targets only 50-year-old women. If anyone else sees or clicks on your ad, it will poison the data, and you won’t have reliable numbers to work with.
How to Make it Happen
Step 1: Create a page on your website to send people to.
You’re not trying to get people to visit your website. Clicking is simply the way they vote on the ad, and they have to click somewhere, so you can simply use your homepage.
But since people are already going to be clicking over to your website, if you want to be a little more advanced, you can send them to a landing page to sign up to get the book or join your newsletter.
In a pinch, you can just use your homepage.
Step 2: Go to Facebook for Business and click “Create Ad.”
Facebook puts this button in several places, but make sure you click “Create Ad.” Do not “Boost Post.” Boost Post takes you to another area where nothing is right.
Step 3: Select “Get More Website Visitors”
Once you click “Create Ad,” Facebook will ask, “What do you want this ad to do?”
Do you want more Facebook likes? Do you want more engagement on your page?
Select the option that says, “Get more website visitors.” That will allow you to send people to a page on your website.
Step 4: Create your ad using Facebook’s Ad Editor.
The Facebook Ad Editor will ask you to upload the photo you want to use in the ad. Every ad needs a photo. Even if you’re testing the title, you still need a photo. If you don’t have a book cover yet, you can use a stock photo or a photo of yourself. It doesn’t really matter as long as it’s the same in both ads and somewhat relevant.
Step 5: Pick an audience with at least 25,000 people and no more than a million.
I like to build an audience based on people who like similar things. If I write books like James L. Rubart, I want to target people who like his Facebook Page. Some authors have targetable audiences, and some don’t. It’s hit and miss, and there doesn’t seem to be a rhyme or reason.
The goal is to choose a focused, tight audience related to your book, made up of as many people as possible. The best rule of thumb is to target the fans of a bestselling author in your category. Build that audience until it reaches 25,000 people but no more than a million. Once you have more than a million people, you’ll be outside of your specific audience, and that will skew your data. Your target audience won’t be more than a million.
Step 6: Click Publish.
You’ve added the image, the title, and the blurb. You’ve picked your audience. Next, you click “Publish.” Your ad is now live!
Step 7: Open the Facebook Ads Manager
You’ll see your single ad in your dashboard.
Step 8: Select the ad you just created a click “Duplicate.”
Duplicate that ad, so you have two copies. Now you have two copies of the same ad running simultaneously.
Step 9: Click “Edit” and change the one thing you want to test. (Book Cover, Title, or Blurb)
Click to “Edit” the second ad and change the one element you are testing. This is when you replace the image with your second cover or replace the title with your second option.
Resist the temptation is to change several elements at a time. If you change more than one element, you won’t know which thing caused the difference in clicks, and you won’t get reliable data.
It’s normally cheaper to show your ad to a large audience. Facebook will show your ads to hundreds of thousands of people, and you will start to get data on which ad got more clicks. You want at least ten clicks for each ad before you know you have good data.
The more money you spend, the more people will see the ad. The more people who see it, the more clicks you’ll receive. The more click you get, the more confident you can be that your data is statistically relevant.
Whatever you do, it will be better than surveying your friends.
I have done these tests, and I’m always personally disappointed because the greatest number of people do not like the ad I like. I’ve had to come to grips with the fact that my personal preferences have no bearing on whether a thing will be popular.
Realize that your personal opinion is probably just as invalid as mine. Test on a bigger audience to check. Use this method of creating an ad, duplicating it, and changing one element to see what resonates with the audience.
How much do I need to spend?
Plan to spend at least $25 per ad, which means in this A/B test example, you will spend $50. If you wanted to test three titles, test Title A against Title B first. Then if Title B wins, do a second test for another $50 total to test Title B against Title C. If you want to test more than three titles, you can create a bracket system where Title A and B run against each other and Title B and C run against each other. The winners then go head-to-head.
You can spend a lot of money going crazy with testing if you test lots of things. I recommend testing only the title and cover. The blurb is harder because people don’t pay as much attention to the text in the ad. You need a larger budget, and the ad must show to more people to get good data on the blurb.
Split testing does cost money but using the data you glean may be the difference between a book that sells hundreds of copies and a book that sells thousands of copies.
While these ads probably won’t pay for themselves, they can benefit your brand. People are learning that you exist. They may not remember your name, but when they see it again, they’ll have a vague recollection that they were once interested in something you made. It may be just enough to get them interested in your book with its scientifically tested title and cover.
MyBookTable is a way to quickly and easily build an online bookstore on your WordPress website. Use it to rank #1 on Google for your book and to boost your book sales on sites like Amazon and Barnes and Noble. Get it free at MyBookTable.com. Novel Marketing Patrons save 25% on MyBookTable Pro. www.mybooktable.com.
Nerdy and pretty good. I wouldn’t mind you going one step farther and comparing several real-life examples of A-B tests that you conduct with us, your Frankenstein monsters. It may take a month or two to collect the examples, but would be worthwhile.