(AllAboutROIMag.com Exclusive: Download the Chi-Square Test spreadsheet to prove your tests are statisically valid.)
If you're wondering, "How am I going to increase my online sales and profits in 2010?" here's an easy answer: TEST! An e-commerce site should run at least one test a month with the potential to increase sales by 10 percent. These are seven keys to a successful online testing program:
1. Test Items That Matter
Create tests in areas on your site where you have a lot of shoppers and sales. You can't test everything, but here are a few areas that matter:
- checkout process;
- search engine marketing and search engine optimization landing pages;
- email sign-ups;
- homepage navigation; and
- cart page.
2. Test Randomly
Split test on your site. Don't run one side of a test for two weeks, then the other side for the following two weeks. Too many things in the environment can change and alter the outcome.
3. Test Significant
Quantities
There are two components to statistical significance: the number of respondents on each side of the test and the number of people in the test. The lower the response rate, the larger quantity needed for a statistically significant result. Many marketers, however, reach conclusions and end tests before they're statistically significant. The problem is that random chance may be responsible for what looks like a winner.
4. Test Significant Metrics
Only test response metrics that can produce a statistically significant result. If you have 50,000 email addresses and a typical email campaign produces 50 orders, you don't have enough quantity for an email creative test that uses the number of orders as the "response" metric. You won't get enough respondents on either side of the test.
5. Have a Plan
Tests should be a critical part of your annual marketing plan. They can produce "easy money" opportunities for online merchants. Create a full year of tests, complete with who's responsible for executing them by when. Review results quarterly.