Facebook Pixel
Join our Facebook Community

How to Use Multivariate Testing to Build the Ultimate Opt-in Form

Posted By Guest Blogger 30th of January 2013 Blogging Tools and Services 0 Comments

This guest post is by Adam Connell of Bloggingwizard.com.

There’s a testing technique out there that’s not being used to its full potential—or even used at all by most website owners.

Today I want to show you how you can use it to create the ultimate high-converting opt-in form.

So what is multivariate testing? It’s essentially very similar to split testing. The difference is that it takes into account a lot more variables.

Many site owners avoid multivariate testing as it seems overly complex, and most of the services on the market that provide multivariate testing are paid services, which leaves bloggers unsure of the potential ROI.

In this post you will learn how you can use Google Analytics content experiments to conduct multivariate testing on your own opt-in forms in an easy and controlled way that will allow you to maximise your conversions.

Why multivariate testing?

In early 2012 Econsultancy.com and Redeye conducted a survey http://econsultancy.com/uk/reports/conversion-rate-optimization-report that yielded some interesting results.

Multivariate testing came out as the most valuable testing method for improving conversions, despite only 17% of companies stating that they used it.

According to the same report, taking the leap from A/B split testing to multivariate testing can help you improve conversions by an extra 15%.

This shows a huge opportunity for those site owners and businesses that come on board and start using this testing method.

So let’s see how it’s done.

Step 1. Break down your opt-in form

In order to conduct any worthwhile experiment you need a plan and identify all of the different variables; but in order to come up with a complete list of variables you need to break your opt-in form into its various elements.

Here is a combination of the typical elements you may find in an opt-in form:

  • headline
  • subheadline
  • additional text
  • image/video
  • name capture field
  • email capture field
  • buttons
  • background.

Step 2. Define your variables

Now that we have all of the elements of your opt-in form mapped out, we need to break each element down further and plan out how we might want to vary each one.

Please note, the list below is not exhaustive, nor do you have to vary all of these when you come to experiment. The point is to show you all of the possibilities.

You may think some of these are minor changes, and they are. But the impact of some of these changes can be enormous.

For example, some marketers have tested opt-ins with name capture and email fields against forms with just an email capture field, and have managed to increase conversions by 20%. So it all makes a difference!

  • Headline: font, text size, text colour, capitalisation, alignment
  • Sub-headline: font, text size, text colour, capitalisation, alignment
  • Additional text: yes/no, font, text size, text colour, capitalisation, alignment, bullet points
  • Image/video: yes/no, image size, image content, video size, video content, video audio, video type
  • Name capture: yes/no, text in field, icon to the left
  • Email capture: icon to the left, text in field
  • Button: size, shape, text colour, text font, text size, background colour, rounded edges
  • Background: border, image, drop shadow, border.

Step 3. Plan the test

This is where it starts to get a little bit more complex: you need to come up with the original control version of the form for your test, and as large a number of variations as possible.

The downside to Google Analytics content experiments is that you’re limited to nine variations plus the original (or control) version.

You also need to be able to keep track of the variations and changes that you’re making; you can’t just throw something in and hope for the best.

To make this easy for you, we’ve put together a Google docs spreadsheet that will allow you to keep track of all your elements and variations.

Click here to access the spreadsheet

Please note: you must make a copy of this spreadsheet before altering it, otherwise everyone who visits will be able to see your testing plan!

testing-tracker

Due to the number of variations that may be needed in the future we’ve broken the document down into controlled groups.

Now just add the variations, which may look something like this:

testing-tracker-filled-in

At this stage it’s important that you only fill in the variations for group A as you need to use the results of each group’s test to inform the variations you select for the next group.

Step 4. Gear up to test group A

Now that you have planned out your variations for group A, you’re ready to get the test underway.

The test

The setup process here is fairly straightforward:

    1. Set up a new page for each variation.
    2. Add the pages to Google analytics content experiments. Log in to your account, then navigate to standard reporting > content > experiments.
    3. Set your goals. Note: the easiest way to do this is to ensure your opt-in form directs users to a thank you page, then find the URL and add this as the goal URL.
    4. Add the content experiments code to your pages.
    5. Let the experiment run.

It’s important to let your experiments run for as long as possible, so you can get data from the largest possible amount of traffic.

The more traffic you run this experiment on, the better, but if your blog doesn’t have as much traffic, then you will need to run it for even longer.

You are just looking for conversion rate here so, strictly speaking, you can run each test on different numbers of traffic. You need a statistically significant result for each test; you don’t need every test to involve the same amount of traffic.

Step 5. Review results and prepare to test group B

By now you will have had the results from group A, which means you can start thinking about the group B tests.

The first thing to do is to take the best performing variation from group A and add it as the original for group B (don’t forget to update your main page on your website at this point).

Now it’s just a case of rinsing and repeating the process above, tweaking and coming up with new variations to test each time.

A potential 15% conversion boost

Using this guide you will be able to create additional experiments for other parts of your sites, not just opt-in forms. You can easily tweak this method to use on sales pages, product reviews, squeeze pages, ad layouts or anything else you can think of.

The important thing is laying out your variations and keeping track of them. Then, just rinse and repeat.

Are you using any form of testing at the moment? We would love to hear about which methods you’re using and how much you’ve managed to increase your conversions in the comments.

Adam Connell is an internet marketing and SEO nut from the UK. He can be found blogging over at Bloggingwizard.com. Follow him on Twitter @adamjayc.

About Guest Blogger
This post was written by a guest contributor. Please see their details in the post above.
Comments
  1. “some marketers have tested opt-ins with name capture and email fields against forms with just an email capture field” – is it true?

    • Hi Johny, thanks for your comment.

      Yep, it’s true – in the studies I’ve done and seen, actually dropping the name field altogether can improve conversions.

      Marketing is all about reducing friction between yourself and your customers, even just that seemingly small field can help reduce that friction.

      The only trade off is that the emails you send to your list won’t be as personable.

  2. Great tips Adam, thanks.
    I think multi-variate testing is going to get much more popular over the next couple of years as people realise how effective it is…. Best of all though, if your testing improves your bounce rate (which it often does) you get an SEO boost too.

    Better conversions + extra traffic = big increase in profit…
    It’s hard to understand why everyone isn’t doing this already.

    • Mark, my pleasure and I really appreciate you taking the time to drop a comment!

      I completely agree with you – it’s surprising how much of a difference that SEO boost can make too.

      I think part of the reason why so many people aren’t doing it is because people have been put off initially by assumptions of how complex it might be. Hopefully now that there are a growing number of tools to help with this and guides like this that utilise the free tools that are available we’ll see more of it.

      Thanks for your comment!

  3. Hi Adam,
    Excellent post, I have done multi-variate testing in past but i did not organize it as you have arranged every thing in a sequence. you have given a brilliant way of arranging the whole information in Google docs. I will try it now :)
    Thank you !
    John

    • Hey John,

      Thanks for the kind words and dropping a comment, I really appreciate it.

      Glad to hear that you’ve been working on some multi-variate testing yourself. I really hope the way I’ve organised this process helps you get more out of optimising for conversions.

      Give me a shout on twitter and let me know how you get on with it (@adamjayc).

      All the best,
      Adam

      • Hi Adam,

        yes i have started testing your method, I will wait and check for few weeks and see how optimizing for conversions go and then will let you know on twitter. :)
        Thank you

        • Hey John,

          Glad to hear you’ve already started using the method from my post – look forward to hearing from you with your results.

  4. Hello Adam, great post!
    I’ve done some multi variant testing but not as documented as this. I recently read an article showing an increase up to 30% just by removing a field which wasn’t really required.

    Even the color makes it change. I guess the more traffic you have, the more interesting results you can get.

  5. Hey Servando, thanks for commenting!

    I think I read the same article, it was about optimising a contact form I think right?

    You’ve highlighted a great point there – you’ve got to have traffic .. the bigger pool of data that you have the better so if you have lower traffic then it’s worth spending longer on testing.

    It’s surprising the difference changing a single element can make, it’s just a case of finding what works.

  6. Thank you, Adam for true and interesting story ;)

  7. Using this guide you will be able to create additional experiments for other parts of your sites… maybe-maybe. Thanks a lot, Adam for this article, very interesting!

    • I’m glad you found the article interesting Carl and I appreciate you dropping a comment. Let me know how it works for you.

  8. Let’s do it again, Adam ! Thanks for good mood ;)

A Practical Podcast… to Help You Build a Better Blog

The ProBlogger Podcast

A Practical Podcast…

Close
Open