Scenario:
Passware’s Windows Key application is a well-known product for making easy work of a very common problem; resetting Windows passwords. Passwords may be lost or forgotten, or the person who knows the password no longer works with the company. When this happens, the results can be devastating, and too many individuals and businesses alike find themselves in the position of having to reformat their hard drives, reinstall Windows and give up on lost data.
With a long-established product, Passware were already receiving extremely healthy volumes of traffic to the product page. And while sales for the product were already excellent, server log analysis revealed that there was room for improvement in the page’s performance.
Strategy:
Rather than creating a series of page designs, then allowing sufficient time to generate reliable data, Passware decided to engage SoftwarePromotions to run a series of multivariate tests on the product page. This allowed us to experiment with far more page variations in far less time; effectively letting us generate more data and better results in two months than we might have achieved in a whole year of manual tests.
Results:
The effects achieved were impressive – with an overall improvement of 34%.
The first test involved identifying three sections of the web page that we felt demonstrated the greatest potential for improvement. Three new variations were created for the page headline, three for the first paragraph of text, and three for the main feature list. The test therefore utilised a total of 48 different combinations that were displayed to visitors. As the first test is basically exploratory, it can sometimes produce little by the way of tangible results. So both we and the client were delighted to see the best variations produce statistically accurate improvements of 29% and 35%:
The second experiment was considerably quicker, with the aim of testing some of the results obtained from the first. We wanted to be certain of some of the results obtained before engaging in the third and final experiment.
The third and final experiment applied all the data gleaned from the initial experiments. We setup a total of 60 variations to be tested, then waited for the information to come in. As we felt that we had already amassed enough information, we correctly assumed that this would be the final experiment for this web page. We allowed sufficient time to generate all the data that we needed, and several good contenders were revealed.
With almost 3,000 conversions generated by the top ten variations alone, we saw the top performing combination generate a healthy and accurate 530 conversions; an improvement of 34%:
After three different tests, we settled on the combination that produced a 34% improvement to the number of conversions. Passware applied the changes to their page, and we moved on to optimising the next product page. Watch this space for the next success story!
