How to Quickly Triple your Click-Thru-Rate (CTR)
September 30th, 2009 | Author: Dan Harrison
In an attempt to boost my commissions via affiliates on my eco-gadget review site EnviroGadget, I wanted to see if I could improve the number of visitors clicking through to the merchant website. I don’t get a commission for all of the products I review on EnviroGadget. However, for the products I do earn a commission, I wanted to make sure that I was getting as many people clicking through my affiliate links as possible.
Therefore I decided to run some experiments with different types of button. Just by changing the colour of a button and changing its text a little, I found a way to triple my click-through rates (CTRs)! Yep, you read that right, triple! Read on to find out how.
The Technology
I have some relatively complex code that I’ve written to handle the affiliate tracking, and it’s beyond the scope of this article to explain it in detail. However, I’ll cover the basics. My affiliate tracking code allows me to create a URL like this:
http://www.envirogadget.com/recommends/a-solar-powered-gadget/g_bn
When that URL is clicked, it redirects the user to the product page on the merchants website, dropping a cookie to ensure that if the user makes a purchase on that website, I get a commission.
The ‘a-solar-powered-gadget‘ part of the URL is the code for a particular product. The ‘recommends‘ part of the URL tells a script on EnviroGadget to redirect a user to a product page using the product code I just mentioned.
The ‘g_bn‘ part of the URL is an example of a special tag that I can use for additional tracking. So I could link to a product using the URL above using many different coloured buttons. If I used a different tag for each different button, it means I can track how many times each button can get clicked. This is what I used to test the different buttons in the experiments below.
Experiment 1 – The Hypothesis
All my affiliate buttons started off being a green Buy Now button. However, I wondered if a contrasting button colour would out-perform my green button (especially as the rest of the site is themed with green). I also wanted to test different button texts too, just too see if the wording had any effect.
Hypothesis 1:
“I expect contrasting blue buttons to outperform the more subtle orange buttons on click through rates”.
Hypothesis 2:
“I expect that the less imperative More Info buttons would outperform the stronger Buy Now buttons”.
Experiment 1 – The Results Data
| Button | Clicks | Percentage of Clicks |
|---|---|---|
| Orange – Shop Now | 141 | 30% |
| Orange – Buy Now | 135 | 28% |
| Blue – Shop Now | 62 | 13% |
| Blue – Buy Now | 58 | 12% |
| Orange – More Info | 53 | 11% |
| Blue – More Info | 26 | 5% |
Button Experiment 1 - Orange and Blue
Notes: I was told that “Shop Now” was a good button text to use, so I wanted to test that. For some reason, I didn’t use green in the first experiment. Can’t remember why!
Experiment 1 – Conclusion
From the results, it’s clear that my initial theories were completely wrong!
So the results show that the orange buttons considerably outperformed the blue buttons. The orange Shop Now button more than doubles the CTR compared to the blue button. Perhaps this has something to do with the emotive effect of colours, e.g. warm v.s. cold colours. Even still, the orange buttons are harder to see with a green background, and blue is more of a contrast to green.
The results show that More Info is a waste of time, performing the worst overall. However, Shop Now is the strongest performing text, with Buy Now being a close second.
Based on these results, I realised I needed to test some more combinations as my assumptions were very wrong.
Experiment 2 – The Hypothesis
Following on from the first experiment, I thought it would be wise to compare the performance of my existing green Buy Now button against orange buttons. I also wanted to test red against orange and green to see if it had any significant impact.
Hypothesis 3:
“I expect that the brighter red buttons will outperform the green and orange buttons, with the red Shop Now button being the best button for the highest CTR “.
Experiment 2 – The Results Data
| Button | Clicks | Percentage of Clicks |
|---|---|---|
| Red – Shop Now | 81 | 21% |
| Red – Buy Now | 78 | 20% |
| Orange – Buy Now | 50 | 13% |
| Orange – Shop Now | 50 | 13% |
| Red – Buy This | 37 | 10% |
| Green – Shop Now | 28 | 7% |
| Green – Buy Now | 27 | 7% |
| Orange – Buy This | 25 | 7% |
| Green – Buy This | 7 | 2% |
Button Test 2 - Red, Orange and Green
Experiment 2 – Conclusion
I had clearly learnt something from the first experiment, and my theory was spot on correct in the second experiment. However, the range of values I obtained was very interesting.
The results showed that using a red Shop Now button (21% of clicks), rather than a green Buy Now button (7% of clicks), there’s a 3-fold increase in clicks through to the merchant website!
In general, red outperforms orange and green. The Buy Now and Shop Now are strong call-to-actions, which resulted in a high CTR compared to Buy This. I’ve not been able to come up with a reasonable explanation for the poor performance of the Buy This button text.
Overall Conclusion
From running these experiments, I learnt 3 things. These 3 lessons are:
- Check your assumptions – I was making incorrect assumptions, so testing my assumptions really paid off. I spotted my mistakes.
- Experiment and track everything – When you do test something, collect as much data as possible so you can analyse it.
- No-one will do it for you – I had to run my own experiments to learn my own lessons. Don’t rely on my data either, test your sites yourself too!
I hope you enjoyed the article and that it motivated you to run your own tests. If you like the pretty graphs, please leave a nice comment, as they took me ages to create!
Update – 30th Sept 2009
John Andrews wrote a great complement to this article on Always Be Link Building. John quite rightly emphasises the point that button colours, and therefore their respective click-through rates, are very specific to a website and its design. So just to re-iterate, test different button combinations yourself, don’t rely on my data alone!
Posted in 
Tags: 
Fantastic post Dan!
I spend too much time on trying to get more visitors when I should be focussing on converting the visitors I am getting.
Inspiring post, so I’m gonna give all this colour changing buttons malark a go :)
Mark
Thanks Mark. Took a lot of work to get the data.
From what I’ve learnt recently, you need a balance of new blood and decent conversions. No point getting new traffic if you’re not converting though.
The next test I need to run is the effect on conversions But that’s going to be much more difficult to test! :)
Dan
Very interesting post – great to see some real examples of data rather than the theory that lots of people tend to post!
Graham
Thanks Graham. The evidence shows that theories can easily be wrong too!
A few people told me that when I mentioned it, it sounded interesting. That’s why I decided to share my findings.
Dan
While the experiments are great, I think that one aspect is the TOTALLY contrasting color to the rest of your site, hence the buttons REALLY stand out.
It would be interesting if you were to change the basic colours on your site to say, orange and repeat the experiment.
I think then you may find that the blue and green would outperform.
Are you up to it?
Matt,
I certainly know how to do it, but I don’t want to. I’ve spent a long time getting a consistent branding for EnviroGadget, so not willing to change the colouring. The whole point of the green is to give visitors an idea of what the site is about.
Any affiliate commissions need to be weighed up against branding and retaining visitors.
Dan
Very interesting article. How long did you trial each combination out for?
Thanks Jon. I ran each test for about 2 weeks (roughly).
Dan
This is certainly a good experiment I’d like to try it myself. Could you give me some hints on how did you do the split testing? Did you display a different button for each visit or for each user? Did you use custom coding or some script? How long do you think it would be good to run the test to get a decent sample?
I also noticed that in the best selling section on the homepage you’re using the “more info” button. Why are you using that button if it didn’t convert well for you?
Hi Javier
It was all a custom script. You can achieve most of the split testing using php’s rand() function. I displayed a different button for each page view. Since I use caching on all my sites, I used AJAX to call my php script, which sent it the HTML to load a new button. The logging was done in a MySQL table.
I ran the tests for as long as it took to show me a significant trend. Statistics taught me to have a set of at least 30 results, but I aimed for that per button to get a real idea of trend.
The new content units are running a new test ;)
Dan
Another aspect is the SIZE of the buttons. They are pretty big. In fact some page designers would almost go as far as saying they are oversized.
This would be another interesting experiment.
Hi Matt,
There are infinite experiments you could run. It’ll never end :)
Dan
Just a thought for your testing – I always see a surge in sales from my sites at the end of each month. Most of what I promote is expensive so I figure people are more inclined to shop on or after their pay day. So if your tests only run for two week periods one period might be in a “hot” pay day period.
Anyway I’m off to change my easycontentunits to red buy now buttons!
Hi Rob
Interesting point about hot periods… however, with an international audience, there could be cultural differences at work too.
With your ECU buttons, make sure you record your results to see which buttons work best. Don’t assume red will work for you!
Dan
A really interesting read, however I often think it’s better to find absolutely brilliant products, may be even unique, often topical products that sell themselves rather than relying on anything other than the product itself to make the sale. Does that make sene? What I’m trying to say is think like a buyer, find products that the buyer already knows they are going to buy whatever the button.
In an ideal world, products that sell themselves would be great. However, in practice, products are not 100% unique as others also try to sell those products. Therefore its about trying to encourage the visitor to buy via my link rather than someone else.
Many of the products I review are already of interest to the reader, so I’m just going the extra step.
Dan
Great tests there Dan and something I haven’t thought about doing myself until now.
I would expect that ‘Buy Now’ and ‘Shop Now’ buttons would perform much better in terms of conversions as people will only click these if they are looking to purchase. A ‘More Info’ button may be clicked by people who aren’t actually intending to buy.
Despite the increase in CTR, did you notice an increase in buyed items?
Thanks Darren. I’m running more tests soon!
Javier – a slight increase, yes. However, my conversions are pretty poor, so not enough data to test yet. My poor conversions are the motivation for me testing new ways of converting traffic.
Dan
“Despite the increase in CTR, did you notice an increase in buyed items?”
Yes, that was my question as well . . . when dealing with affiliate programs if a person isn’t prepared to buy they aren’t prepared to buy regardless of whether they click or not – although an increased CTR certainly can’t hurt . . .
Regardless, thank you for the post Dan it is certainly food for thought . . . I know when I started my retail site several years ago I read that using “buy now” was not a good idea and that “add to cart” was better but this is the first article I’ve seen that discusses button wording in regards to affiliate sales so I follow it with great interest.
Thanks Again! Karen
Thanks Karen.
I’ve started tracking which products/buttons trigger a sale, but it will be very tough to create any meaningful analysis for a few months.
My advice is, don’t assume, test it!
Dan
Fantastic psychology experiment Dan, I never would have thought that the colour of a button would impact the click rate! Will have to try it our for myself, thanks for the very useful post!
How bizzare, I think I shall be experimenting with differnt coloured adsense ads now!
Thanks
Love your classical hypotheses – null hypotheses approach here Dan. It took me back to my university psychology days. I tell you this would have been a great idea for a study then.
Interesting conclusions here and Dale’s post above prompts a question; do you think that text links would have the same effect? I’m hypothesising no but would be really interested to see what your results are Dale.
Thanks Matt. I thought it was important to explain my train of thought when I was running the tests.
Dale’s point about colours for adsense does have an impact though. Seen some subtle results myself.
Dan
I to am trying some split testing at the moment with different button colours and different call to actions. Hopefully I can get some good ideas what will work/convert best in the run up to christmas.
Yep. although since it’s Christmas, you might need to consider the effect of seasonal influence. Perhaps use plenty of cartoon Christmas banners to draw attention to Christmas gifts and offers?
Nice experiment Dan.
I would say that the “Buy This” would score very low because:
- It is not specific. “This” is usually considered very informal, not credible, and not specific enough to take action (it eliminates the call to action effect)
- It sounds more of a declarative statement than an offer to sell something.
Hi Pinny
Interesting perspective on “Buy This”. I have an alternative theory… namely that visitors feel that it commits them to buying that product, when they still want the opportunity to change their mind.
Dan
Dan,
That is also a definite possibility – great idea :-). We area society all about choice and infinite selection right now….