The biggest mantra in the conversion world is “test”. A lot of people focus on the possible negatives of testing. I get asked a lot of questions like: “Don’t you worry about hurting conversions?” “What if my test doesn’t improve conversions?” You need to remember that in the end you will improve conversions. It doesn’t always happen on the first try. You’re going to either prove or disprove your hypothesis but gain knowledge in how users interact with your site, and that’s what will help improve your conversions in the long haul.

Now you have to understand that there are four parts to any test. You have data analysis, hypothesis, actual testing, and the analysis of the results. I’m going to quickly break down each of these categories so that we can gain a better feel and understanding for testing along with why and how it should be done. In all honesty, each one of these can be a post in itself. (I feel a mini post series brewing)

First off, you don’t just come up with random things to test. That doesn’t help anything or anyone. You have to start with data. You need to look at everything from traffic to how a user interacts with a site or page (there are great tools out there for helping with both, an article will follow soon with a break down on some of them). The goal of looking at all this data is to learn/find where the leaky holes are.

So we found the leaky holes but now what? You need to build your hypothesis and it better consist of more then just turning the submit button from blue to red. Consider who’s coming to the site, what there intentions are, and what the leaky holes tell you. You may realize that people are spending time in the wrong parts of your page or site. You may see that people are clicking on things that aren’t links. The goal is to try and come up with changes that will benefit the user and you.

Let the testing begin. You now have a separate version with the changes to test your hypothesis. Set up your test environment, split the traffic and let it ride. Now the amount of time you let s test run really depends on the amount of traffic your site sees on a regular basis. You need to have enough data to make sure there isn’t any fluke traffic causing spikes or downward trends in your conversions. Unless you see thousands of unique visitors a day, I would let the test run for 5 days, and then see what the numbers say. Just remember in all test, the environment needs to be as neutral as possible.

Now before you look at the results and start digging into what worked and what didn’t, you need to remember what you’re doing here. The overall goal of conversion and testing is to get a better understanding for the person using your site. You either proved or disproved your hypothesis, and in turn gained a better insight into your visitor’s behavior.  Overall, every test you run will bring you closer to improving that conversion rate.

.

.

.