| By Rick Cook
A/B Testing, Multivariate Testing and Test Marketing are a Continuous Journey
The most successful marketers are the ones constantly testing the responses to their marketing campaigns. Testing results in incremental improvements over sustained periods of time which then drives better campaigns with tighter focus and improved response rates. Yet most companies don't test their marketing campaigns. They simply launch them and watch what happens. According to Mark Jeffery in “Data Driven Marketing” a recent survey of top companies showed that more than two-thirds of them did not conduct test marketing with control groups before launching campaigns. Below are some tips and tricks to add value and measurable success to your marketing campaigns.
Use CRM Software To Keep It Under Control
The reason most companies don’t test market is that it takes time and effort to manage multiple campaigns. A firm with an active test marketing system may be running 12 or more campaigns at once. If you try to do it manually, or with spreadsheets, the process quickly becomes laborious and prone to error.
Customer Relationship Management (CRM) software systems, and more specialized lead management systems, provide systemic methods to automate much of the effort, making test marketing much more manageable. With CRM applications you can cull out your test lists, prepare your campaigns and monitor the results with much less effort. Marketers managing multi-channel campaigns and desiring automation with deep analytics are clearly going to be best served with marketing automation systems, such as Infor's CRM which combines CRM with marketing automation in a single package.
Set Your Goal
Obviously you want your test campaign to be a success. But how do you define success.
Decide what metric or metrics will determine the success of your marketing campaign. Will it be Return On Marketing Investment (ROMI)? Response rate? Leads generated? Sales produced? Response rate is the simplest metric, but it is considered the least useful for most purposes. ROMI is a measure of profitability, but it can be harder to figure and it isn't applicable to all campaigns. The point is you have choices and you want to choose wisely. This is especially true since you'll probably want to use the same success factor for most or all of your marketing campaigns.
Make Sure Your Test Sample Is Representative
While there are very sophisticated techniques to make sure you are generating as close to a random sample as possible (true randomness is very hard to achieve) generally in marketing for small and midsize businesses (SMBs) it is enough to have a representative sample. When choosing your sample try to avoid anything that would bias the results. For example if you choose, say, every 10th name on the list, make sure you draw your sample from the entire list. If every 10th name is too many, take every 20th or whatever. But make sure your test sample is evenly distributed.
Don't Try Too Large A Sample
A test sample needs to be large enough to show a response but not much larger. A sample consisting of half your list – unless it's a very small list – is a waste of resources.
How big is big enough is a matter of judgment. For large lists (10,000 names or more) a sample of 500 is more than adequate, as long as it is well chosen and representative of the larger list.
The key thing is that the sample needs to be large enough to show results. A too-small sample may mask the actual response to the campaign in background noise and show no apparent response to your effort.
Keep It Simple
Only try to test one thing at a time. Multiple criteria and variables muddy test results.
Start With Major Variables
Generally there are three major variables in any campaign: Audience, Offer and Message. When you start a campaign on a product or service, try to vary one of those three things. You can fine-tune the details later.
Include A Control Group
A control group is a sample that doesn't receive the marketing campaign. Control groups help you establish a base line – what your customer activity would be without the campaign.
Test More Than Your Message
Once you’ve run your initial tests on Audience, Offer and Message, you can run additional tests to model cause and effect impact and then fine tune your response for the next iteration.
Even the smallest things can have consequences. For example, one company that ran a heavy schedule of direct mail advertising campaigns discovered that it got a higher response when the mail was sent using stamps rather than a postage meter. And that they got an even higher response when the stamp was put on the envelope upside down. Why it happened was inconclusive but it happened and they took advantage of it.
Similarly, it is understood wisdom in email marketing that the best time to send a campaign to businesses is Tuesday and Wednesday mornings about 9:00 a.m. The best time to send a campaign to individuals is about 5:00 p.m. Friday. The theory is that businesses are less distracted and more receptive early in the week once Monday's backlog is out of the way. Individuals, on the other hand, are more likely to be responsive at the beginning of the weekend. How true this might be for you may be another matter, but many email marketers report different results depending on the day and time of day they send out their messages.
Testing is a continuous process. You should constantly run test campaigns to measure the response to your efforts. One test campaign is better than no tests, but repeated tests are better yet. Always assume you can do better and work toward that goal.
Tags: Test Marketing, Control Groups
Author: Rick Cook