2. Execution
Now develop a list of variables to test. If you use an email service provider, it may be able to offer some valuable insight on quick wins you can initiate easily. Don't forget you will need to test everything twice to be certain of your results. A great strategy is to identify one stream of communications -- for example, your regular monthly newsletter -- and splitit into two versions. Whichever comes out on top, use that as the control for your next test. Testing should be a continuous activity, and by using this method you can ensure your marketing is always evolving.
Here are some thoughts on what you might want to start testing.
Subject lines: First thing first: You need people to open your emails if you want any hope of selling to them. You probably already suspect which aspects might resonate with your customer base -- try including price, sale deadlines, and unique variables such as loyalty points to see what piques their interest at a glance.
Content: You may always lead with a strong hero image at the top of your communications, but maybe it would be better to let the content float higher up. Is there value in segmenting based on geographic location and sending people content based on where they live? Perhaps your emails are very short and inspirational but would drive more qualified traffic if you included more detail.
Demographic segmentation: Do you always use bright pink buttons to compel people to your website? Perhaps male customers would respond better to a royal blue button, or maybe there is value to be gained from showing different products to different demographics.
3. Reporting
There are two key aspects to explore with regard to reporting: where to be looking, and how to find a winner.
You need to know which success metric is the most relevant to your test; there will always be a primary and a secondary. The subject line example we looked at in the introduction is the perfect example: Our primary success metric is the open rate, and this is really the only thing a subject line can hope to affect. Pay close attention to your unique click-through rate as a percentage of open rate, not just volume. If you have a great open rate but an awful unique click-through rate, you will know that your subject line is attracting the wrong type of people.
Content is a little more complex. You may develop an email that extols perfectly the virtue of a new product, proven by a record unique click-through rate. If you then find that these clicks are not converting to sales, something on the web page must have put them off, right? It's easy to blame the web team, but you might want to ask yourself some tough questions too. For example, if you got this great unique click-through rate by omitting the price from your creative, then your results are of very limited use.
Both of these examples consider the primary metric as being the action closest to the variable, and the secondary metric as the next action after that. This is really as much as you can hope to have influence over; any action further removed from your variable has been subject to too many other factors. You can't really attribute high conversion rates to a great subject line.
Lastly, it's essential to make sure your answers are statistically significant. Even two equally sized random splits will give slightly different response rates, so it's not enough to look at the highest value and declare it the winner. You will need to do some calculations to make this decision.
4. Take action
Testing initiatives can become very involved, so don't lose sight of the goal. Once you've got your first wave of results, make sure you share them with everyone involved and take action on them. But the job doesn't end there. Monitor your results to ensure your new best practices stay fresh and are still producing the desired results. Again, a good testing plan never ends. Your customer base never stops evolving and neither should your marketing plan.
Tim Underwood is an account manager at e-Dialog.
On Twitter? Follow iMedia Connection at @iMediaTweet.
<< Previous page
