If you’re like most marketers, you’ve been there.

You have a brilliant idea for your next A/B test. You’re absolutely positive the B version is going to be a smash hit. Your conversion rate will go through the roof, you’ll become Marketer of the Year, and you’ll bask in the glory.

But when the results come in, it turns out your B version hasn’t moved the needle at all.

It’s a disappointing situation, but you’re not alone. A Google search for “A/B test ideas” will turn up over 85 million results with all kinds of suggestions. You could try making your landing page longer, putting your copy in bold, changing the color of your buttons, adding a video, or any number of other things.

But how can you know what will actually make an impact on your conversion rate before you spend the resources on another A/B test? It’s impossible to try everything. Besides, what works for one company won’t necessarily work for the next. (If there was one perfect solution for improving conversions, we’d all be doing it already!)

The truth is, it’s hard to know what’s going on in your customers’ heads. You don’t know what they’re going to click until they actually click it—and even then, you can’t be sure why. As marketers, we suffer from the curse of knowledge: we look at our websites every day, so we can no longer see them with fresh eyes. And what seems clear and simple to us may be quite the opposite for a first-time visitor.

The only way to really know what your users are thinking, and what they want to see on your site, is to ask them directly.

Here are three ideas for generating better A/B tests—with the help of your users:

1. Watch a user go through the normal flow on your website.

Run a simple user test, either by inviting customers into your office and watching over their shoulder, or by setting up a remote usability study with a tool like UserTesting. Have the user start on your homepage and try to do what a customer would normally do on your site.

For an ecommerce site, ask them to find an item, add it to their cart, and go as far as they can through the checkout process. For a SaaS company, have them learn about your features and figure out how to set up an account or request a trial.

Make note of places where they get lost or stuck, or where they miss details that seem obvious to you. Did they ignore the call to action? Maybe it’s time to try changing the position and size of the button. Did it take them a long time to figure out what you’re selling? Try tweaking your copy. You’ll come up with a list of A/B test ideas that are much more likely to make a difference than testing randomly.

2. Survey your site visitors to find out why they’re leaving.

It’s very easy to create an exit survey on your site with a tool like Qualaroo. You can set up your survey to appear when a visitor leaves an important page, like your checkout page. Ask why they decided not to purchase today, and find out what they would change about the site if they could.

3. Find out the why behind your past A/B tests.

Think back to the last A/B test you ran. You know which version won, but do you know why? Is it because of the color, the copy, the layout, or something else altogether? If your B version has several different variables, figuring out the answer to this question can be especially confusing.

Run a user test, like the type described in tip 1, and show users both versions. Then ask which version they prefer and why. You might find that it was easier for them to understand the copy on the B version, but the bright button on the A version caught their attention more. You can use these insights to come up with a new version that combines the best features from each of the previous ones.

When it comes to A/B testing, you don’t want to make guesses and hope to stumble onto the next big idea. A/B tests can be a time-consuming process, meaning you’ll never have an opportunity to try every idea out there. So make sure you’re setting yourself up to make an impact with the ones you do run.

Once you find out what your users are thinking, it’ll be much easier to come up with a winning B version. Rather than a wild guess or a gut feeling; it’s a direct response to feedback from your customers. You’ll get to the bottom of the real issues on your site so you can start improving your conversion rates. And then you can bask in the glory.

Have you tried getting customer feedback to get ideas for your next A/B test? What’s the most successful A/B test you’ve run?

Running with idea image courtesy of Shutterstock.