How to Use Multivariate Testing to Perfect Your Marketing Strategy

When incorporating testing into your marketing, it’s important to understand your limitations.

There’s no shortage of advice on designing a marketing strategy that works. One thing that I wish someone had told me before I ever built my first marketing strategy is the fact that this advice is only worth as much as you can implement, measure, and test.

Now, testing your marketing is an interesting challenge because there’s no standard definition as to what constitutes a “test.” What one marketer considers a “test,” another marketer in the same industry may write off. On the other hand, some tests that more advanced marketers run wouldn’t even be recognizable to their less experienced counterparts.

When incorporating testing into your marketing, it’s important to understand your limitations. Plenty of marketers aren’t ready to jump into multivariate testing, and that’s fine! For those marketers, here are a few other options.

Different Ways to Test Your Marketing

If you’re looking for some more basic approaches to test your marketing, try the following:

  • Basic Experiments: A basic experiment is the most simple test you can run with your marketing. When running a basic experiment, all you’re doing is trying something new in earnest for a statistically significant period of time, I recommend 1-2 months. A good example of a basic experiment would be changing the “From” email on your marketing emails from your company to an individual on your team.
  • Platform Tests: When experimenting with a new platform, a basic experiment isn’t enough. If you want to find out if a new social network is a good fit for your brand, you need to commit to testing it for at least 3-6 months. This includes things like Twitter, Facebook, Email and blogging.
  • A/B/C/D/E Tests: This is where things get interesting. Once you’re ready to start diving into large-scale testing,you can experiment with A/B testing. This means that you can send an email with two different subjects to two similar lists to see which performs best. This is the baby step that leads to multivariate testing.

How to Use Multivariate Testing in Emails

Once you’re ready to start multivariate testing, you’ll never want to stop. Unfortunately, most major marketing automation platforms don’t have multivariate testing built-in. This means that you need to do it yourself. Honestly, it’s not as hard as it sounds, just do the following:

  • Export your email list as a .csv and open it in Excel.
  • Add a new column A.
  • Type in “=RANDBETWEEN(1,1000000)”, fill the entire row.
  • Sort Column A smallest to largest.
  • Delete Column A.
  • Divide this list as you see fit. If you’re testing 5 variations, split it into 5 lists, etc.
  • Upload these lists into the software you use to manage your email.
  • Send a unique variation of your email to each of these lists.

A couple important notes here. Number one, it’s important that you have a “baseline” for your test, so have one email that is your “standard operating email.” This should be testing nothing, it’s just an email like any other you’d send. Once you have this baseline, only change one thing about it per test email.

Once you hit send, you can measure your results by list to understand what variation is most effective. For best results, try 2-3 times per list to ensure one isn’t abnormally responsive.

How to Use Multivariate Testing in Digital Ads

Digital ads can be vexing for a lot of marketers because there’s so little to work with. By limiting headline copy to 25 characters and putting strict requirements on imagery, Facebook and Google seem to get a kick out of eliminating creativity from ads.

That being said, this can make digital ads one of the most effective mediums for multivariate testing. What I do with all of my digital ads is create 5 headlines, 5 bodies and 5 images. Then, I mix and match every permutation of these elements until I have over 100 ads. That allows me to test each variable independently. It’s not glamorous or easy, but the results are phenomenal.

How to Use Multivariate Testing on Your Website

If you’re committing time and money driving people to your website, you might want to start testing the pages people land on to ensure you’re not missing any opportunities once they arrive. Personally, I like to use Optimizely on all of my headlines and calls to action in order to get a better understanding of what language resonates with my target market.

Subscribe to
The Spark!

Join brand managers everywhere getting smarter about marketing each week.

Oops! There was an error sending the email, please try again.

Awesome! Now check your inbox and click the link to confirm your subscription.