How a Landing Page with the Call-to-Action Below the Fold Outperformed an Above-the-Fold Variant by 304%

Since you didn’t specifically ab test the CTA location, it’s hard to describe the test the way you did. I agree that it’s plausible the early CTA was too aggressive, but it’s not conclusive. Especially for a lead gen offer, where typically short copy can do really well, even for complex offers.

I’d challenge you to cut the offer down to a free 30 day trial, 1 headline, 2-3 bullets MAX, and your same form. Then see what happens..
p.s. I tried typing my name in lower caps but it didn’t work.

  • Michael Aagaard says:

    Hi Gab – thanks for the comment.

    In this case, the objective was to challenge the best practice “Always place the CTA above the fold.” The research question was: Which variant will before better – the control /CTA above the fold) or the treatment (CTA below the fold).
    The test design was a simple A/B split with a 50/50 distribution between variants. The sample size was 100 conversions (conversions not visits), the statistical confidence level was 98%, and the standard error was <1%.

    The result was that the treatment outperformed the control by 304%, and the answer to the research question was “The treatment (CTA below the fold) performed better than the control (CTA above the fold).”

    I’d venture to that all the abovementioned facts provide the basis for a pretty conclusive test…

    Testing a variant 30-day trial on a short-form landing page is interesting. However, that would be a completely different test and we’d no longer be challenging the “Above the fold” myth, we’d be testing whether a short-form LP performs better than a long-form LP, or whether a free trial performs better than an introduction offer. Thus we would also have to come up with a different research question.

    But funny you should mention it, because the first thing I did for this client was to challenge their original short-form LP that consisted of a headline, a few bullets, and a paragraph. My long-form LP more than doubled conversions…

    The free 30-day trial is an interesting approach but – as mentioned in the post – the product in question is a physical service where you get ingredients and recipes delivered to your door. The costs involved with the supply chain and delivering the product make it difficult to justify a free 30-day trial. Thus the company has chosen to go with a special introduction offer involving a discount.

    Thanks for the heads up about the caps, it’s setup by default in the theme I’m using, but I’ll look into changing it.

    Thanks!

    - Michael

  • Michael,
    Congratulations on a huge increase in conversions! It’s easy to read this and think, “Oh, I guess in every case, we need to put the CTA below the fold!” But, as you noted, “find out what works in your specific case, on your particular target audience.” Do you feel like women were main TA and they felt having the CTA at the top was too pushy? Just curious. Thanks for the article!

    • Michael Aagaard says:

      Hi Darren – thank you very much!

      Good point – what a backfire if I actually ended up inventing yet another lame, generic best practice ;-)
      Like you said – The goal here is to inspire people to test what works on their specific page instead of relying blindly on best practice.

      I don’t think the results had much to do with the demographics. My hypothesis is that, because of the complex nature of the offer, it’s way to early to ask for a conversion the very second someone lands on the page from a PPC add.

      Dr. Flint from MarketingExperiments uses the analogie that asking too much too soon on a web page is like walking up to a girl in a bar and asking her to marry you before she’s even had a chance to find out the first thing about you ;-)

      Thanks for reading – hope you stop by again soon!

      - Michael

  • Hey Michael – I love that you had the idea to ‘swing for the fences,’ this is generally where ideas for many of the best variations come from.

    I will say that I don’t believe you can attribute you massive lift to the CTA location. If you boil down all of the attributes on the page that were changed:

    1. The pictures of the dishes that customers would receive recipes for
    2. Moving the testimonials from a vertical stack to a horizontal span
    3. Making the social proof with 40% call-out both centered and significantly larger
    4. Changing the positioning of the grocery images from vertical left-aligned to horizontal full-width
    5. Adding additional graphics, button, and call-out to conversion offer above the sign-up box

    There is really no way to attribute this lift solely to the location change of the conversion form. This is something I actually explained in relative depth in a recent presentation on making sure you use real (and accurate) data both for test planning, design, and especially success attribution. Feel free to check it out if you like: http://www.slideshare.net/nickeubanks1/split-testing-using-real-data

    I don’t mean to come off negative at all – I’m honestly thrilled that people are beginning to challenge what have been toted as ‘best practices’ which very well may have evolved beyond their legacy applications.

    Thanks again for the post.

    • Michael Aagaard says:

      Hi Nick – thanks for your comment!

      I agree – in fact, in the post, I make a point of mentioning that there were other variables in the test that would have affected the result as well.

      The point of this case study was to challenge the myth of the “CTA above the fold”, and the fact remains that the variant with the CTA placed below the fold all the way at the bottom of the page significantly outperformed the variant with the CTA placed above the fold a the very top of the page.

      Thanks for reading and commenting – I appreciate your feedback and hope that you’ll stop by again soon ;-)

      - Michael

    • I was going to basically say what Nick said here. There’s a lot going on here, as you noted, but I think it’s quite significant that you broke the left margin with the CTA. I do love that you’re challenging some common assumptions, and I would love to see some additional testing.

      (Related: http://blog.kissmetrics.com/shocking-truth-about-graphics/)

      • Michael Aagaard says:

        Hi Carson – thanks for your comment!

        The main thing I’ve learned from split testing is that human beings and decision making processes can be pretty unpredictable. Which makes a strong case for being scientific and challenging the common assumptions.

        I have a back catalogue of 200+ split tests and more coming in every week – so stay tuned for many, many more upcoming case studies!

        - Michael

  • Proof that possibly ‘small’ changes can make HUGE differences. Split testing should never end in this business

  • Great experiment! I was thinking for the two sites I am working on to place a CTA on 2 spots (example: on blog post on the top right corner and bottom of post), but this experiment is now guiding me into different direction. Thanks for sharing.

    • Michael Aagaard says:

      Hi Chande – my pleasure, glad I could help!

      There are a lot more experiments and case studies on the way ;-)

      - Michael

  • Hi Michael,

    The comments above already made the first point I was going to do.
    But there’s one more thing you A/B test… I think if you double the CTA either to above and below the fold, you could see some interesting results.

    Let me know

    • Michael Aagaard says:

      Hi Alexandre – thanks for your comment!

      Yes, that’s an interesting idea, I’ve considered it myself.

      However, I don’t work with the client anymore, so I can’t test it.

      - Michael

  • Great stuff Michael. I like that you quoted Flint in your comments. Perhaps adding that additional context to the blog post would help people relate better to your primary thesis here.. that the landing page is a conversation…. and you don’t start a conversation by saying “Buy this now and I’ll tell you why you should later”..

    PS. Love the way you show your results just like our boys at M.E….
    Cheers

    • Michael Aagaard says:

      Hey Matt – thank you very much! I’m glad to hear you like my new blog!

      Great idea with the Flint quote – I’ll have to incorporate that!

      Stop by again soon ;-)

      - Michael

  • Amazing post Michael – I wish everyone would have this outlook!

  • I think I could do a similar test with my newsletter subscription section to see how does that perform below the fold once readers are done with reading the article on post page. What say?

    My site is designingtips.com and I’ve recently just been focusing on adding more and regular content to it, less conversions & monetization.

  • This article is really interesting because common practice have always taught us to put optin box and CTA boxes above the fold.

    But based on what you’ve found, it depends on the complexity of the content. If the content is easy to digest then put the content above the fold. But if it’s too difficult and technical to comprehend, below the fold would increase my conversions for my CTA boxes, right?

    • Michael Aagaard says:

      Hi Ryan – Yes, that’s exactly my experience.

      But my main point is that you have find out what the best placement is on your page. And the only way to do that is to test it in real life on your potential customers.

      - Michael

  • great to see someone actually do it differently. Everyone religiously follow what others say and wont try things like putting the CTA in other places other then the right side or in this case at the bottom.

    Great post!

  • What if we add both top and bottom CTAs? Is this overkill in your eyes?

    • Michael Aagaard says:

      Hi Benny – Thanks for your comment.

      It’s difficult to say without having seen the landing page. It depends on the individual case.

      If you send me a URL or a screen dump, I’d be happy to give you my thoughts. My email is: michael@contentverve.com

      - Michael

  • interesting read, i have started playing with my layouts but never really considered moving the cpa cheers ;)

  • hey Michael,

    you have a very interesting case study here, congrats!

    now, I feel the need to answer this suggestion:
    “Your call-to-action should always be above the fold!”
    not to prove it wrong, but show when it’s valid…

    Yes, you need to have your call-to-action above the fold in any of these situations:

    1. When you’re doing a pre-launch to your mailing list or to an audience who has read about you/your product before — and all they care about is the order button…

    2. When you’re focusing on “instant gratification” buyers or opt-in mailing list subscribers, who dislike scrolling, and just want the goodies right there, right then!

    3. When you care about numbers – you prefer quantity rather than quality of the prospect/client.

    There’s no CTA case study I can share with you, as that’s not what I do online, but my findings were from personal testings and what I’ve seen others do on the Internet: how people buy and click, etc.

    Hope it helps

  • Hi Michael,

    Challenging the current thinking can only ever be a good thing! Always keen to read about new ideas.

    However, I don’t think this test necessarily proves that CTAs are not always better above the fold – essentially there are too many uncontrolled variables. Your hypothesis might very well be true, and I would be very happy to see it tested in isolation, but the different page designs above can only contribute to further testing, forming two outliers within which to design further.

    To be honest, I suspect that what you have actually proved in the above test is that engaging the user is vitally important. The Control is visually unengaging above the fold – extremely text heavy, no imagery, few interactions, with an uneven vertical modular layout. It is not surprising the page converted poorly. The Bounce rate and Dwell time of the page would be able to quantitatively establish this.

    Conversely, the Variant has bright, colourful and relevant imagery in exactly the right location, there is a clear and neat modular structure, which allows users to visually process and digest information faster, and there is a single encompassing title. I suspect if you were to test placing a well-designed CTA above the fold within this new page, you might find no difference between that and your original Variant – this would imply that the main fault of the Control was not the placement of the CTA but the page content itself, and that this is what you have corrected. Performing a before and after comparison of page dwell times and exit rates would lend further visibility.

    All in all, this test is a good foray into the unknown, and an interesting challenge to the general thinking on orientation around the page fold, but rarely can such any single dramatic redesign prove beyond doubt the fallibility of such a fundamental tenet of design and optimisation standards – it simply provides a signpost for further exploration.

    • Michael Aagaard says:

      Hi Ollie – Thanks for your comment!

      Yes, there are certainly other things that have been changed on the page – I also made a point out of mentioning this fact in the case study. Nevertheless, the fact remains that the variant with the CTA at the very bottom of the page outperformed the control version with the CTA at the very top of the page significantly – something that should not be possible if you subscribe to the Best Practice Rule that above the fold CTA is always better. That’s the main point of the case study.

      As for your point that the control is “extremely text heavy”; there is exactly the same amount of copy on both variants, and the copy is identical on both variants – the only difference is the font size of the header and sub-header (It’s smaller in the Control).

      Moreover, I’ve conducted several other tests since then where variations with the CTA below the fold have outperformed variations with the CTA above the fold.
      I’ve also conducted several tests where variations with the CTA above the fold have outperformed variations with the CTA below the fold. So – just to make it clear – I’m by no means saying that the CTA should always be under the fold (that would be stupid). What I’m suggesting is – like I state in the article – the optimal placement of the CTA depends on the product being sold, and the motivation of the prospects – and not a standardized “cookie-cutter” rule of design best practice ;-)

      Again thanks for you insights – I appreciate you taking the time to interact!

      - Michael

  • It’s really a cool and useful piece of info. I am happy that you simply shared this helpful info with us. Please stay us informed like this. Thanks for sharing.

  • Hello Michael,
    Like yourself I am a great believer in testing then testing again and again. In fact I do not stop testing on any individual item that can affect website traffic or sales. Over a long time I have discovered that placing calls to action in positions that are not generally recommended sometimes give results like you mention. I have also found repeating the same call to action in two places on some pages is also a winner.
    Toby.

  • Great experiment ! It’s a useful info. Thanks for sharing.

  • While researching information about CTA placement I found a lot of people referred to this post. However, I can only find comments about the post, not the post itself. Is it still available?

  • Speak Your Mind

    *