DAY #14: GETTING STARTED WITH A/B SPLIT TESTING
When it comes to increasing conversions, it doesn’t pay to guess. Website owners often make changes based on hunch or intuition. For instance, they might think a blue banner would work better than the green one, and so they make the change. Yet, they have no real way of knowing if the blue one was objectively better than the green.
The only way to know for sure is to test both versions. A/B testing allows you to do that. Generally, you test the current version (the control) against the new version. You take the one that wins, then you pick another test and you repeat. This way, you can consistently improve your web pages over time.
Testing Philosophies: Fast Data vs. Statistical Significance
If you ask a mathematician – or a programmer – how much data you need in a split test, they’d likely give you a fairly complicated answer. You need enough data to establish a conversion range for both versions. For example, A converts at 3.5% to 4.5% and B converts at 3.7% to 5.2%. On the surface, this looks like B is better than A. But it’s also entirely possible that B actually converts at 4% while A converts at 4.5%.
Split testing tools will calculate the range of possibilities for how well a page converts. Then, to have true statistical confidence you need to wait until both ranges do no touch. In other words, one range has to be clearly above the other with no overlap. Finally, you have to wait until you have at least 99% confidence in that range.
This is a fairly complicated, and a fairly lengthy process. That’s why most online marketers take a much shorter and faster approach. They wait until one performs better, then they choose that option and move on. They don’t wait to know how much better it is; nor do they wait for the full set of data. They make fast decisions.
In business, you can’t always wait for complete mathematical certainty. To wait for mathematical certainty, you might have to wait one month to do one test. Using the right tools can help speed up this process by guaranteeing that you know what visitors are doing on your site..
Test Big, Then Narrow
Start by testing big ideas. Try adding a photo if there isn’t one, or add a video. Try completely different selling angles. Try completely different sales language. Try using a phone number instead of a form. Try a different price, or try using a payment plan.
Test big first. Once you’ve tested a few big ideas and found the ones you want to stick with, you can gradually test narrower elements. Always test one thing at a time; otherwise you won’t know what exactly made the difference.
Finally, always be testing. If you’re getting traffic to your website, you should absolutely be using that traffic to test your data. Even incremental improvements on a daily and weekly basis can result in massive boosts over the long run.
Today’s assignment like many of the previous ones is on-going. You’ll start it today but will need to continue to work on it over the coming weeks and months.
- Make a list of the big things on your website that you want to test
- Create a secondary list of the narrow items
- Pick one big item and create a split test
What We’ve Covered
Day #1: Examine Your Audience and Your Core Offer
Day #2: Optimizing Your Call to Actions
Day #3: Improve Your Lead Forms and Increase Signups
Day #4: Your Headline – Your Page’s Most Important Sentence
Day #5: Keep Attention With Strong Transitions
Day #6: Web Design Tips for Better Conversions
Day #7: Reduce Shopping Cart Abandonment by Improving Your Checkout
Day #8: Boost Conversions by Managing Expectation and Relevance
Day #9: Master the Art of Creating Urgency
Day #10: Use Video to Boost Trust and Conversions
Day #11: Add Social Proof to Your Sales Process
Day #12: Accessibility Makes You Real (and Improves Sales)
Day #13: Tailoring Your Guarantee and Return Policy
Day #14: Getting Started With A/B Split Testing
Coming Up Next
Day #15: More Ideas for Better Conversions