Split Testing Tips For Small Businesses

This post was written by Josh Krafchin of CleverZebo.

It ain’t easy being a growing business who wants to split (A/B) test.

It’s so important to do – it’s how you find out what offers and approaches your audience responds to.

But there’s so little time to learn how. So we tried to answer all the basic questions right here so you can get your answers all at once.

(Incidentally, here’s how to test in AWeber.)

To Get The Most Results, What Should I Test?

When in doubt, start simple. Testing the basics will help you determine the main elements of your campaign. Later, you can keep those elements in place while you tweak the details.

Test Idea #1

If you want to test your email newsletter, try testing plain-text vs more heavily designed HTML emails.

Plain-text email open and click-through rates often rival those of their fancier, graphically designed brethren. If you’d rather a more designed look, templates are available.



Depending on your audience, you may get better response with message-from-Mom-type plain text, or they might prefer a more colorful or visual design.

Once you test it, you’ll know which approach to take.

Test Idea #2

For another testing option, rather than break up a free trial registration into multiple steps, why not test including it all on one page?



You could find that your registrants are less likely to abandon the form after finding they have to fill out a second or third page.

On the other hand, you could find that seeing a fuller one-step form could deter people from even starting to fill it out.

Test Idea #3

Have a product whose image sells itself? Feature it on your sign up form!

Recent studies show that people respond far more to images than text, so any image is going to draw their eyes to your form.

The question is, which image will best encourage subscriptions?

Check out this beautiful (because of its simplicity) landing page:



How Can I Test Despite Low Traffic On My Site?

Probably the biggest factor in determining the scope of a test is traffic volume (followed closely by traffic quality). Without traffic, there’s little to test.

The good news is, even if you have even a small amount of traffic, you can start testing to see what engages those site visitors the most.

Testing For Low Traffic (Fewer Than 10 Daily Visits) And Minimal Analytics History

The absolute first priority, before you start testing, create one best-shot landing page to start with.

Focus all your energy on expressing your value proposition succinctly, creating a powerful call to action, listing convincing benefits and explaining how your product or service delivers those benefits.

Testing is useful for optimization, but unreliable for guidance through the basics of around how to sell. Before you worry about testing, create the best-looking, best-written, most compelling landing page you’re capable of and start there.

After you’ve launched that landing page (or if you already have) let the analytics for that page – reports on how much response you get from each part of the page – guide your next move.

If there’s a part of your page that’s getting clicked on less than you think it should (especially in comparison with other parts of the page), start there.

If I Have More Traffic, What Should I Test?

Testing For Moderate Traffic (100+ Daily Visits Or At Least 3,000 Past Visits)

If your analytics are showing high engagement and conversion rates, go to the next section. Otherwise, it’s back to the drawing board.

You may have a deeper issue than messaging or design. Start your creative process over.

Perhaps it’s a conversion funnel problem (Offer a free trial? Give away a free resource in exchange for contact info?)

Perhaps you were speaking to the wrong audience and need to re-imagine the concept altogether.

Regardless, don’t worry about small changes at this point – go for the whole enchilada. You need to find an approach that converts and a small change like a different button color probably won’t do it.

Testing For Heavy Traffic (Thousands Of Daily Visits)

If your landing page is converting poorly, see the previous section. Otherwise, since your site is bringing in traffic, we’ll assume your basic elements are good. Now it’s time to get more minute with your testing.

Elements to test, in loose order of importance:

Keep in mind, the first and last elements of your page that people see are the ones that will stick out most in their minds (meaning, the ones that will have the most influence on whether or not they click to the next place your page leads them).

How Can I Make Sure To Get Clear Results?

What’s wrong with this test?

VS.

In case the suspense is killing you, B generated 40% more sales. The problem is that it’s not clear what about B generated the sales.

Was it because, nearing Valentine’s Day, any Valentines-related offer would win?

Perhaps it was the different copy and presentation of the shipping language?

Or maybe it was the color difference, pictures, and picture placement?

While Smiley Cookie found a winning landing page, it’s guesswork to know why version B won. To get clear results, they (and you) would need to test one thing at a time, then use the winning version to test a different change.

To be fair, Smiley Cookie seems to be a fairly active conversion testing company, and this is only one of many tests, but it’s a good example of how promising landing page tests can be undermined by testing too many elements at once.

What Tools Can I Use To Run These Tests?

In the world of landing page testing, there are many options; here are a few of the more prominent ones.

Unbounce is a great tool for folks without technical or design backgrounds who want to get landing pages and landing page tests up quickly via a WYSIWYG editor. Optimizely is a great option for marketers who want to quickly change elements on a page without bringing in a developer.

The key is to find solution at a price point and with a functionality set that works for you.

Here’s a shortcut for you – if you find a piece of copy that works in one channels, say a headline in PPC or email subject line, leverage it in another area – your landing page headline, for example.

Ultimately, the goal is to find the right combination of target demographics, traffic sources, conversion funnel elements and lead nurture components.

Bonus Suggestion: Take A Risk

With so much at stake, it’s easy to get overly serious when working on a marketing program.

Don’t.

Especially if fun is what will win the day.

This is the full-screen, image-focused landing page for Spotify (a music site!):



When you think about it, it doesn’t make sense that big beautiful images of trees would sell a music service, but the end effect works, and it’s so simple:

Image, call to action, headline.

In a dozen words, they’ve told you everything you need to know to sign up.

Brilliant.

Josh is the founder of Clever Zebo, an online marketing consulting firm. When not developing cutting-edge online marketing strategies and responding to email, Josh can usually be found taking walks in the park or working on his jump shot.

Print This Post

Related Articles

Become a Better Email Marketer

Subscribe to This Blog by Email
Why Subscribe?

5 Comments

  1. Hi Josh,

    Super points here. I am already split testing away. But it’s great to get inspired with views and ideas.

    When split testing stuff it’s rare to guess a winner upfront!
    My users / customers and visitors keeps me guessing! That’s the fun (or revelation).
    Before I make a statement: I do a split test :)

    6/18/2013 5:04 pm
  2. Thanks for the advice, i designed a new marketing idea from the above info within an hour & everyone loved it. Now to monitor the replies :)

    6/18/2013 9:57 pm
  3. I also think that testing email subject lines is a good way to split test and make your email campaigns profitable. But i do like the idea of testing “text” vs “html”… this is something that i preach to my email subscribers.

    6/19/2013 11:53 am
  4. Peter, good to hear you’re already split-testing. And really good to hear you’re observing/listening to your customers. That’s the most important part.

    Tanya, what’s the marketing idea? Mind showing us? Would love to see it.

    Nice, Randall. Great stuff.

    6/21/2013 12:08 pm
  5. I still haven’t tried split testing but your post and example of Spotify helped me edit my sign up page so it has less words and a new image.

    7/10/2013 12:12 pm

Leave a Comment

Need an avatar?



Follow Comments
Send me notifications of new comments
rss Follow comments via RSS