Website Testing Wins: Orange or Blue?

Website Testing Wins: Orange or Blue?

By Peter Borden | Jan 30, 2012 at 9:20 am

There’s been a lot of talk about how orange buttons increase conversion. Orange buttons were said to increase conversion by 32% in one case study and big orange buttons have been said to boost conversion in blog post after blog post.

But, if you dig into the results, you’ll see that orange buttons were almost always tested against a control of no button at all. In cases like these, it’s hardly a surprise that orange buttons make a difference. Of course they do … when compared to no button at all! Practically any button will make a difference, regardless of color.

So, we wondered, do orange buttons really make a difference? What if we compared orange buttons to, say, blue buttons?

Well, the results might surprise you.

One of our clients recently tested the two buttons below on their product detail pages:

The blue button won. In this simple A/B test, we found that the blue button outperformed the orange one, producing a 9% lift.

So much for that bit of online marketing dogma. Yet again, this just goes to show you that you always have to test.

Peter Borden is the marketing manager at Monetate and is responsible for PPC and email marketing strategy. He is the author of the weekly Website Testing Wins series every Monday. Peter's also an expert on the psychology of persuasion, influence, and conversion as well as a developer of iPhone apps prior to working at Monetate.

http://twitter.com/pavlicko david pavlicko

Yep, you gotta test this one. My hunch is whichever one is the contrasting color to the page layout.
- Peter Borden
  
  Yeah, David, that’s what I would have thought as well. Unfortunately, I can’t release the full screenshot because the client wants to keep that confidential. But, I’ll tell you this: while the blue *is* contrasting, it’s not nearly as much of a contrast as you might think. Frankly, I would’ve expected the orange to perform better myself. Just goes to show you that you always have to test!
http://www.MarketAgent007.com/ David Padilla

I tested an orange button against an equally-bright yellow (gold-ish) button. The orange button won. I would like to test the orange against a bright blue button.
- Anonymous
  
  A good rule to testing is: test big variations first, small variations later. So, I’d always suggest testing two vastly different colors, and then narrowing in on the right hue.
  
  But that assumes I’m testing button colors in the first place. Which is a rarity.
  
  In fact, we tried to dissuade this client from testing button colors at all.
  
  Why?
  
  Because fiddling with button colors is one of the least efficient things you can do to improve conversion rates on your site. There are MUCH quicker and larger wins to be had. Read our 10 Best Practices Ebook and you’ll see what I mean: http://monetate.com/ebook/10-testing-targeting-and-personalization-best-practices/
http://www.conversionista.se/ Chief Conversionista

Colour is meaningless. Contrast isn’t.

I always tell my clients that the most important thing about the main CTA btutton is that it STANDS OUT against other elements.

If you’re not showing in what context the button appeared, it’s kind of pointless.
- Anonymous
  
  I can tell you this: the main color scheme of the page was red menus on a white background. While one might imagine orange blending into the overall red/white color scheme, I didn’t see it that way … I felt both the orange and and blue were equally high contrast.
  
  What’s more, color DOES matter … more than you think … it isn’t “meaningless.”
  
  Over the last decade or so, we’ve seen a sort of visual language emerge, with different elements, and different colors, meaning different things. This “visual language” has been pushed forward by the sites with the largest audience — sites like Amazon, eBay, etc.
  
  As these sites have adopted consistent color schemes and UX element usage — blue underlined links, for example, and orange buttons — consumers have come to associate these common usages with meaning.
  
  This is why it’s so common to see orange buttons associated with checkout, and blue underlined text with links … and, of course, to see increased conversion associated with using these common elements.
  
  The test summarized in this blog post, therefore, was a pleasant surprise insofar as it deviated from what most people have come to expect.
Pingback: 50 Ways To Seduce Your Web Visitors With Persuasive Landing Pages
Pingback: 50 Ways To Seduce Your Web Visitors With Persuasive Landing Pages | Sphinx Web Design Experts
Pingback: figurative - diseño gráfico, websites y experiencias, centrado en el usuario | Flesh Art Fair: carne de arte con piel de diseño y comunicación
james lofton

It’s not just the contrast of the button to the background, it’s also contrast of the text on the button. The blue button’s text possibly stands out more on the deeper blue back ground. Run an accessibility color validation test or simply screen shot and convert to monochrome to check/compare the contrast of the elements.

There could also be bias if the test was not limited to only the same product. Having the buttons appear across a group of product pages, even with random selection, you still run the risk that the blue buttons were appearing on pages for products that had more competitive pricing.

Lastly there could be cultural and even gender bias introduced if the site has a predominant audience type that favors one color over another. Which would then skew the results just for that site and make these findings applicable to only the site and audience it serves.
- PeterBorden
  
  About the biases you mention:
  
  You’re right in theory, but not in practice.
  
  The whole point of having a large enough sample size (and of getting statistical significance) is to render such randomness meaningless (or relatively inconsequential) to your results.
  
  The reason tests require thousands and thousands of visitors before reaching a conclusion is because of such random variation. But, given a large enough sample size, variations of gender, or the pricing of products that these buttons appear on, become inconsequential to the results, leaving only the impact being made by the variable you’re testing.
  
  For more, here are two good places to start:
  
  http://explorable.com/statistical-significance-sample-size.html
  http://parkinslot.blogspot.com/2003/12/sample-size-and-statistical.html
  
  That said, you’re correct about test results only being applicable to the particular website and the audience it serves. NONE of the website testing wins we feature on this site should be taken as universal. You always have to be testing. In the end, it’s the methodology that matters.
Pingback: Which Color Converts The Best? « Nine Seconds
Pingback: Copywriting : Cara Memicu konsumen Menekan Tombol ‘ Buy Now’

Connect with Us

Subscribe to the Marketing Agility Report

Connect with Us

Thanks for subscribing to the Marketing Agility Report

Want to Learn More?

Request a Demo

Thank You!

Website Testing Wins: Orange or Blue?

Products

Solutions

Why Monetate?

Resources

About Us

Policies

Contact