How Changing 1 Word in the Call-to-action Generated a 38.26% Lift in Conversions

Your call-to-action copy is just as important as the button itself, and even minor changes can have significant impact on your conversion rate. This case study is a perfect example of how little it sometimes takes to get a major lift. By changing one word in the call-to-action, I was able to increase conversions by 38.26% on a B2B site.

Background

The client has asked to be anonymous, but I can tell you that it’s a Danish client I work with on a regular basis.

They have a portal through which businesses can find offices for rent. The site features thousands of offices that potential customers can browse through. Once a prospect finds a relevant office, they have to click to the main call-to-action (located on all pages) in order to get more information on the lease sent via e-mail.

This means that clicking the CTA is the main conversion goal, and every extra click potentially means money in the bank.

The Test

I had already conducted a number of experiments – focusing on button color and shape – that led to a dramatic increase in conversions. With the button itself in place, it was time to experiment with copy.

After 4 years of testing CTA copy, my main observation is that the more value you can convey via your button copy, the more conversions you’ll get. I decided to test this theory in its purest form by changing the copy from “Order information” to “Get information”.

The hypothesis being that “Order” emphasizes what you have to do – instead of what you’re going to get. Whereas “Get” conveys value as it emphasizes what you’re going to get – rather than what you have to do to get it.

I set the test up as a simple A/B with a 50/50 split between variations and let it run till we reached a sample size of 195 conversions and a statistical confidence level of 99%.

Here’s an example of an office page and the two variations of the button copy:

The translation of the Danish CTA copy is located right under the button image.

The Results

The Treatment outperformed the Control variant by 38.26%. The button is featured across thousands of pages, so imagine the accumulated impact and potential ROI of changing this one word.

Main Takeaways

I may be repeating myself here, but button copy is just as important as the button itself. Your call-to-action represents the tipping point between bounce and conversion. And in that last critical last moment where your potential customers have to make up their minds, the copy itself is what they’re going to interact with.

Tipping point between bounce and conversion

Over the years I’ve performed dozens and dozens of experiments split testing button copy, and I’ve consistently achieved lifts of anywhere from 5 – to 200% by simply tweaking the words.

Learn to write call-to-action copy that converts

I recently posted a 6-minute how-to video on writing high-impact button copy. In the video I’ve distilled 4 years of research into 1 simple optimization principle that anyone can use to get more clicks and conversions. Watch the video and learn about CTA copy here >>

Comments

  1. Lasse Rubin Skov says:

    Hi there Michael

    I love this sentence and I also have test-data that backs it up:

    “The hypothesis being that “Order” emphasizes what you have to do – instead of what you’re going to get. Whereas “Get” conveys value as it emphasizes what you’re going to get – rather than what you have to do to get it.”

    Still, I was wondering if you also measured the order conversion rate after people clicked the button?

    I was thinking that the “Get information” button could be a bit misleading, for people who might think the “information” was free. The “Order information” button might get less clicks, but might convert to orders better since the clicks could be more relevant.

    I guess its finding the sweetspot between to two :)

    • Michael Aagaard says:

      Hi Lasse – thanks for your comment! I’m glad you liked the case study.

      Actually, getting the information on the lease is free ;-) I forgot to mention that in the post.
      The “Get” Treatment did outperform the Control version on all leves though.

      - Michael

  2. Order information: “If you click here, you will have to pay for the information you want to receive.”

    Get information: “If you click here, you will receive information.”

    It’s an open door if you ask me. Nevertheless, wise lesson learnt. These statistics via A/B testing are really valuable but what would have been more valuable would be to know WHY users would or would not click the button via a user test.

    Perhaps it would be a good idea to test several actual designs at once before publishing, instead of trial-and-erroring (and having to wait on results and thus possibly lose call to actions) on a live site?

    • Michael Aagaard says:

      Hi Matthijs – thanks for your comment.

      I agree, the more “perfect” your website is before launch, the better…

      In this case however, I was hired to optimize the performance of an already existing website. Thus, I had zero influence on what went into building the website.

      User testing is a valuable tool, however, it has its limitations. This A/B test for instance, is based on a sample size of 10.000+ users. It would be pretty difficult, time-consuming, and not least expensive to get 10.000 subjects to user test your website. As far as I know, the standard user-testing sample size is around 10 users. So this means that you’re assuming that the opinions of those 10 people is representative of all future users. Personally, I’d rather base my business critical decisions on the sample size that is 1.000 times larger ;-)

      Moreover, user testing is an “unnatural” test environment. The subjects know that they are taking part in a test, and that they are subject to scrutiny and analysis. In an A/B test, the users don’t know that they are taking part in an experiment. This means that their decisions are 100% natural and unbiased as they are not framed by a question that like: “If you were to visit this site because you were actually interested in the product, which button would you rather click A or B?”

      I prefer a combination of user testing and A/B testing. Start out with user test before launch, and then A/B test whether your hypotheses work on a sample size that’s large enough to give you statistically significant data.

      - Michael

      • Too bad they didn’t ask you beforehand. I don’t know how much it has ‘cost’ them to wait until the A/B tests results were in. Nevertheless, their designer made a boo-boo when they typed “Order information” – you must admit, it’s an open door.

        I am of course aware of the less natural environment during a usability test. However, the outcome of these tests are severely influenced by the (social) quality of the tester and the way the questions are formulated.

        If you would actually need 1.000 times more testers, it might be a good idea to double check whether your client’s product focus is sharp enough.

        • Michael Aagaard says:

          Hi again Matthijs

          User testing is a great tool for picking up boo-boos. Split testing is a great tool for finding out how to best mitigate those boo boos.

          Optimization is an ongoing process – in my mind you should never stop testing your site and working to improve its performance.

          Having user tested a site does guarantee that your site is “perfect”. I’ve often been able to increase conversions on sites that have been heavily user tested before launch. In fact, just an hour ago I received test results from one of my clients showing that my new treatment improved conversions by 15,50% (99% statistical confidence)on a payment page that had been user tested less than 6 months ago.

          I’m a huge fan of both user testing and split testing. As I see it, the two go hand in hand.

          - Michael

  3. OK, Michael Great Tips. What you suggest for a Pricing table button. Think I have 3 different packages for a logo design service. Will the “Order Now!” text on the button is good, or something else you suggest?

    Waiting for your reply.

    Thanks
    Ifham khan

    • Michael Aagaard says:

      Hi Ifham – Thanks for the question. Can you send me a link so I can see the page?
      That way I can give you more qualified advice.

      - Michael

Speak Your Mind

*