In a recent post, I boldly proclaimed that people not doing multi-variate testing on their websites are lazy. To prove that we drink our own kool-aid, I wanted to share the results of some tests we have been running on Linkpatch.com, a web application we built.
Through the use of Google analytics and website optimizer (both are free tools to use), we were able to increase account signups by 74% over a four-month period. Here are the three tests we ran:
1. Re-design the Tour Page
I learned pretty early on that the tour page was not helping us like it should. Early feedback from customers and friends that tested the software proved that the copy was wrong. We needed to explain exactly why we did not create a website crawler or link checker, and clearly outline the unique benefits of Linkpatch over every other available solution.
Some key metrics I was watching in analytics when testing the old tour page versus the new one were "Time on Site", "Bounce Rate" and "Exit Percentage". Here are the results:
Time on Site went up 35% as a result of this change, which is a great improvement in my opinion. Bounces went down 14%, and the Exit Percentage stayed about the same. There is still more to improve on there, but most importantly, the conversion rate for people that viewed this page went from 3.3% to 4.5%.
2. Test Headlines
Next was the main home page headline. A great headline is everything in my opinion, because an effective one will qualify our target audience and explain exactly how the app makes their life easier. Here are the three variations we tried:
Original- "Linkpatch reports broken links and images on your site, and provides all the information you need to fix them quickly."
Alt 1- "Linkpatch monitors your website for 404 errors, and provides all the information you need to fix them quickly."
Alt 2- "Linkpatch monitors your website for errors, so that you can be the first to know when users run into a problem."
As you can see, the differences in copy are subtle. Here's how they did over about a month of testing:
Alt 2 was the variation that proved to be the best headline for us to use. It resulted in a conversion rate of 5.28%, which was 11.6% better than the original we launched with.
3. Test Call to Action Button
The final test we worked on proved to be the most important, because it had the widest margin of difference between variations. Many people have debated the "right" language to use in a CTA button for a web app, and "Plans and Pricing" somehow became the de facto standard for most. However, the reason testing is important is because EVERY site is different. The same rules don't apply 99% of the time, so we decided to test the buttons for ourselves to see what converted best. Here are the buttons we used:
Of the four combinations, the "See our plans and pricing (free trial included)" button won:
As you can see, the winning button had a slight edge over the original one after testing for about a month. What is most important to note though, is the estimated conversion rate by the time the second test was over, compared to the estimated conversion rate in the first test. From beginning to end over those 4 months of testing, that's where we saw a 75% increase in conversions according to analytics:
Although these 3 tests spanned about 4 months, it was all in my spare time. All in all it only took about 10-15 hours of work and study.
Keep at it! Re-work and re-test some other variations to see what works. As the site evolves and things change, there will always be plenty of things you can test on the site in search of a higher conversion rate.
The biggest takeaway for me is to quit fretting about small design and copy decisions prior to launching the site. Give it your best, then create multi-variate tests for anything you feel may impact conversions. There is no way to be right 100% of the time about this stuff, so take the pressure off and your users will tell you what works best.
One final note to keep in mind is that you can't test your website or get great quality feedback until it's launched. We see projects get delayed all the time while internal decision-makers contemplate what they think is best for the website. It should never be up to them if you are in the business of maximizing conversions. Just make an educated guess so you can launch and start testing! That's the only way you can find the most accurate answer.