Remember that science experiment you did back in middle school? The assignment was to grow 2 different flowers. One had to be in sunlight, while the other had to be in a shaded area. The catch was, all of the other variables (i.e. water level and temperature) had to be exactly the same to determine how the sunlight affected the flower’s growth. That's the idea behind A/B testing!
Who knew your crazy middle school science teacher would actually help you learn one of the key elements in the digital marketing world? Read below to learn how to implement tried and true A/B testing techniques that separate average marketers from the top dogs.
Why is A/B testing important for digital marketers?
Essentially, A/B testing is exactly what it sounds like – it’s testing “A” vs. “B.”
In fancier terms, A/B testing brings a scientific methodology to your marketing campaign to understand how small changes can influence customer behavior and make your strategy as effective as possible.
As far as what items you should test, this could be the image you include in your social ad, the headline in your email, the text you use in your call-to-action, or the use of a chat widget on your website. Really anything that can be measured, so start with the campaign component that you believe has the greatest potential to impact campaign results (or lack thereof). Which brings me to my first best practice…
A/B Best Practices
1. Don’t Sweat the Small Stuff
When conducting A/B testing, you should only be focusing on the elements that have the biggest impact your conversion rate. For this step, I like to refer to the 80/20 rule. Find the 20% of your web-pages that generate 80% of your results (a few results examples might include traffic, conversions, lead inquiries, or highest average time spent on the page) and focus on that!
For example, let’s say you discover that the homepage of your website is generating most of your views. But when your reader’s reach the 4th subheadline, they stop reading and click off the page. That subheadline is what you should be testing! Consider changing the verbiage. If the headline is currently written as a question, make it declarative statement instead.
Is your business struggling with inefficient communication processes?
Improve business communications
and eliminate inefficiencies.
OK, we’ve switched from science class references to 8th grade English, but this stuff has huge power to impact results! Or instead of changing the words, change the color of the headline. Or the font. Or the justification from center to left. If the headline is on top of a background image, graphic or white space, try an alternative background. Simple changes can produce big results! There are so many possibilities to try, so make a list and start plugging away. That brings us to our next tip — one that is probably the most important on the list because without it, your A/B testing will be a crap shoot versus a controlled experiment.
2. One at a Time
Don’t get carried away! Be sure to only test one variable at time, so that you closely monitor the effect and impact changing it produces. Your marketing and your campaigns may range from straightforward and simple to intricate and sophisticated, but isolating one testing variable at a time is always the recommended practice. If you’re not seeing the engagement or campaign results you desire, there is likely a multitude of possible reasons why. While it might be tempting to just tear your campaign tactics down the studs and start fresh, or totally revamp, a methodical schedule of testing one variable at a time will deliver clean and clear results. It might take several weeks or even months to get the information you need to improve results, but slow and steady wins the race. It’s called A/B testing not A/B/C/D testing for a reason. And while we’re talking about timing...read on.
3. Even Stevens Across the Board
Before you begin the process of A/B testing, determine how much time you’re going to devote to each testing experiment. Depending on your company and level of audience engagement, you might need to let an experiment carry on for several days or even several weeks to accumulate helpful results. If your website only gets 1,000 visits a month and only 200 of those visits go to your "Service A" page, you’ll probably need longer that a few days to determine if changing the subheadline on the "Service A" page is a beneficial change. Alternatively, if your site gets 20K visits a month and 5K of those visits land on the "Service A" page, you might see the results of your experiment more quickly.
It’s important that you give each A/B test the same amount of time to deliver results. You shouldn’t allow 10 days for one test and 4 weeks for another, because that inconsistency defeats the purpose of testing one isolated variable at a time. Start your A/B testing process with a time frame per test in mind. If you find that the time frame you choose is too long or not long enough, you can modify the time frame for the second test but after that, keep all testing durations even.
4. Use a Testing Checklist
Digital marketers should plan what tests they will be running and why in order to receive an ROI from their efforts. Below is a pre-test checklist to use for an optimal testing process:
- Identify and prioritize your tests: Rank the tests you plan to run by level of potential impact and how much effort they require to implement. Regularly update this list as you come up with new ideas and goals.
- Use a testing calendar to keep your experiments organized: If you’re a one-person A/B testing machine, you might want to keep your process simple, at least while you’re first getting started. Begin with one campaign channel — maybe your Facebook ads or landing page rather than testing multiple areas at once. If you have help or are just a marketing rock star and you are able to conduct multiple A/B tests at once (of course, never testing more than one variable per test at a time), a calendar will be even more critical to keep the tests organized.
- Set goals and record results: You’re conducting these tests for a reason right? Write down the testing goals and closely track the results of your efforts. In other words, Marketing 101.
Make sure you can answer the following questions. If not, maybe you should reconsider if the test is going to be worthwhile.
- What information are you hoping to gain from running this test?
- What variable are you planning to change?
- Will you be able to apply your learnings from this test to other parts of your marketing campaign in the future?
- Is the area you are going to test on significant?
5. Apples vs. Oranges
If you’re unsure where or how to begin, a bold way to begin your A/B testing process is to test two completely different variations of your website to discover which version performs better — kind of like comparing apples vs. oranges. From there, you can conduct smaller A/B tests based on whichever version you found to be more successful.
Creating dramatic variations between your tests will help you get to the root of your problem and maybe even help you uncover new improvements that you hadn’t thought of before. An example of this is using two complete different homepage layouts, or two completely different image/graphic styles on the homepage.
6. Consistency Is Key
Stay consistent with your testing methods. In order to understand if the variables you are testing actually work, you need to expose your test to the same audience, during the same time of day, for the same duration of time.
Think about your middle school flower experiment — it wouldn’t have been effective to grow one flower in the cold winter months and another flower in the summer when warmth and more sunlight impact growth factors. Your marketing experiments are no different. Consistency creates certainty and confidence in your results. The last thing you want is to question the validity of your findings after devoting time and effort to the process.
How To Measure Your Results
If you follow these six testing best practices, you have the tools to run successful A/B tests. Don’t stop at implementation, however! Measuring your results is half the battle and there are plenty of software programs to help you easily measure results. Google Analytics and Optimizely are two of the industry leading options!
HubSpot is a favorite at Stratus Interactive, not only for measuring results, but also for gaining a holistic understanding of your consumer’s behavior. HubSpot offers a built-in A/B testing feature for emails, landing pages, CTAs, and other campaign elements.
So, go ahead and get testing! A/B testing is an essential step in being an effective, modern marketer. You can improve content engagement, reduce bounce rates, increase conversion rates, maximize product visibility, and the list goes on. Have fun and enjoy the process — every piece of data you learn is like a little present from the marketing gods. It doesn’t get much better than that!