How to Run Tests as Part of Your Marketing Efforts

Feb 24

1) Plan ahead

Too many times we’ve seen clients rush into a test without a concrete plan of action, only to find that they can’t adequately track the results of their test. Always plan out the test ahead of time and have a few people involved who can poke holes in the strategy.

2) Have a control group

A lot of things can influence the performance of your marketing. Split your test data in half and have a control group and your variable group. Make sure that you have separate tracking on each, such as different 800#s or landing pages, so that you can say with absolute certainty that a response came from your variable group or your control group.

3) Target the right quantity

Many of our clients, understandably so, don’t want to siphon off a large portion of their data for doing a test. However, without a large group it is difficult to gauge response and conversion. In a group of 100 leads (or even 1,000), one response can make a huge difference in results. We recommend using a minimum of 5,000 per test group (10,000 total between variable and control) to get an accurate picture. That does not mean that you should do all 10,000 in one week if you’re only marketing to 10,000 a week. By doing smaller test and control groups over a longer period of time, you keep from hurting yourself should the test underperform compared to what you are currently doing.

4) Test, test, test

We cannot stress enough how important it is to test your response methods. There is nothing worse than launching a mailing test just to realize half-way through that your phone number isn’t set up correctly and you’re not receiving all of your calls. You should test anything your respondent may try to do and make sure that it won’t derail your test. What if they don’t complete the landing page, or they submit the form twice? What if they call after hours? Before you launch anything, you need to have everything humming like a well-oiled machine.