I have read many articles for the last 3 years regarding behavioral economics and neuroscience and in 2014 I decided to make experiments of my own. I have a small website around 1K unique visitors each month; kartsec.com. This web site recommends you the most advantageous credit card by matching the card’s campaigns with your expectations and your spending behaviors.
The planned business model depended on users filling credit card applicant forms and getting commission on each new credit card customer from the bank.
The site had 3 major features; recommending which card to apply for(wizard1),recommend one of your existing cards for a unique shopping transaction(wizard2), recommend a brand to buy from (wizard3). The first one was going to bring direct income, the others were only ways of gathering information from users to run data mining experiments. So I had to make the first one more attractive than the others.
I started with a basic rule, the left-upper part of the page is the most clicked part. So I placed card recommendation link on that part;
This layout and colors proved that it was right to put the wizard1 to the left-upper part of the page. Wizard1 took 5 times more traffic than wizard2. However there was still more to do here because I needed more traffic going to wizard1. So I decided to change the colors and came up with a second design;
Changing the colors had an incredible effect on the traffic flow. Starting from the main page; wizard1 takes 12 times more traffic than wizard2 and 20 times more traffic than wizard3.
While routing the traffic in the target proportions, I still could make many other experiments to optimize it but one of the most important tasks in behavioral experimenting is to choose the most beneficial part to experiment on. From now on any change would bring small changes on the traffic route so I had to test other things which would bring more benefits. So I decided to work on decreasing the bounce rates, which was about %40, too high for the sustainability of my business model.