The Art of A/B Testing

By all means you need to get people’s email address to initiate any email campaigns before the recipients ultimately become your faithful readers. You need to offer free lunch in exchange for their valuable email address.

How many times do you subscribe to a newsletter to get a bible document or unlock a result to your survey? This is the hook that you can’t resist to.

Now we’re asked to create a cheat sheet for those who are interested in digital marketing. My topic is A/B testing.

If you’re looking for marketing positions in web or mobile based companies, you must know what A/B testing is because it’s a costless but effective way to know your customer and optimize your online service.

A/B testing is a marketing process that measures how two variations of a landing page, web forms, ad, email or other online content are split against each other to determine which results to a hypothetical goal, typically the conversion rate. You can literally test everything to prove your hypothesis. It can be the copywriting, call-to-action, frequency of impression, or even the shape or location of a button. Again, you test to understand your visitor’s behavior and preference while experiencing your online service via PC or mobile device.

Key takeaways of A/B testing after you read this post:

Simplify to one variable in your A/B test with measurable standard to evaluate your result. The goal is to narrow down one specific factor that makes observable change. I’m not a fan of multivariable test. That requires advanced statistical knowledge and skill to analyze the result. And you’ll need extra time and extra data for your test. Save your effort. A/B testing is not an experiment for the thesis of your doctoral degree.

Don’t expect a single test that solves all the problem. A/B test is about incremental gain that moves you towards the optimized online practice. It is a process of continuous improvement and might take several tests to reach an observable change. Test. Test. Test. Most tests will fail. (Some says it’s 1 out of 8.) But you can even learn from failures — user’s behavior online is volatile and you should be on track of every change in the trend. Correct your hypothesis and continue to test.

Be patient. You need to wait until you have statistical confidence to jump to a reliable conclusion. That means you’ll collect considerable unbiased samples enough to substantiate statistical significance level. It’s either data volume more than 500 or a complete cycle of the period you’re testing. A/B testing is actually a social experiment. You should treat it scientifically to claim a confident conclusion. (Remember your science experiment in your old school days?)

If you’re interested in digital marketing, I recommend you to follow PreHack @joinprehack and get updates from the website. You’re not just receiving a free lunch.

Week 3 DigiMktg: #Canva #HTML #CSS #emailetiquette