I tried to use Jakob Nielsen - Don Norman methods of user interface design and testing to improve the website sales of comics. The method of iterative design and test was quick, with about a one day turnaround per iteration. Each cycle was tested with at least three users who tried to order comics (about 10 minutes each user tested) while making notes of what they did. Problems found were immediately fixed and tests were run again.
Testing and user reactions have changed the site in the past, for example, adding a custom programmed comic book search when free search utilities were tested and found to be worse than no search at all. A secure credit card payment page was requested by customers and also helped sales.
Testing the site has showed that almost every tester that used the comic book search would click on a comic book in the search result expecting the action to add the comic book to a "shopcart". Don Norman calls that a "convention" in an affordance and design article. Alas, the site had no shopcart, but had an email ordering procedure.
Email ordering was terrible. Every tester would easily find the email ordering instructions, but most testers would abandon the task of ordering comic books at that point. Some would phone in the orders after writing the choices down. A very few would not give up and actually complete an email order. The order instructions may have been difficult. We did specifically test the instructions and tried to simplify them, but still the usability tests showed the same results, many people would not order as soon as they saw ANY text instructions that mentioned sending email.
So I looked for pre-built free source shopcarts. And tested them. Bah, none fit our business or the customers we serve. Not that these shopcarts did not fit another business model, most were suitable for purchase of one or two items per transaction and a small product domain. Unfortunately, we sell low price back issue comics and to make a profit we have to sell many comics per order, some 10 to 100+ items per order. We also have tens of thousands of titles and hundreds of thousands of issues. If only we sold $5000 gold bars instead of fifty cent used comics we could have used a free shopcart.
We decided to custom program a simple shopcart with the following
- Customer can see and order many items quickly and easily
- Workers have to fill the orders easier than the current "answer the email" system.
- Easy to program, quick to change.
Usability testing showed plenty of problems but most were easily fixed. The biggest problem was that the first version of the shop cart was not accessed from the browse pages, only the search. When testing, people by "convention", still clicked on a comic in the catalog browse page expecting some sort of shopcart feature. Since most of our traffic comes from search engines to a catalog browse page, the disconnect of the browsing and the search based shopcart kept orders lower than they should have been. When I made browsable comics in the catalogs choosable for the shopcart, orders almost doubled again.
Yes, the Norman - Nielsen methods worked, the site does sell comics, more and more with each round of design and test. Are we rich? No, we sell used low grade comics, get real. Are there still improvements to be made on our site? Yes, I am aware of many interesting and wonderful problems currently on the site.