I always get nervous when I hear the word test. It gives me a headache, which in turn empties my brain of everything I know on the subject, rendering me officially in panic mode!
But if only I had more tests like this when I was at school, perhaps I would be less logophobic now as the subject of today’s blog is a test that is not on what you already know but what you don’t and are going to learn.
Now, I don’t think that counts as an oxymoron but maybe you should read on and then put my mind at ease at the end when you are more than welcome to, in fact you’re invited to, give feedback and leave comments…
O.K, now, because I think the word ‘testing’ sounds very clinical and scary, I’m going to put it another way say that in this blog you will learn about ‘experimenting’, or, if you prefer, finding ways to do things better and constantly improving.
Right, so why would you want to run a test in marketing? The purpose of running a test is to determine whether changing methods, variables and components of your communication will have a direct affect on the results of your activity or campaign.
If we never tested things, pushed the limits and boundaries, or changed things we would stand still. All companies and individuals test on a daily basis. Whether it is a new product, a new price, promotion or a new market, we are constantly testing without even being aware of it sometimes. But testing is not just throwing a dart at a dartboard and seeing where it lands, there is a lot more to it and a lot of fundamentals to consider.
Firstly identify the aims of your test, e.g. why are you going to change something and what do you want to prove?
-
Is it to get more sales?
-
Increase average spend?
-
Get more email sign ups?
-
Will it determine whether different channels give different results?
Do you think changing one aspect will bring you closer to this aim as opposed to another? Identify the objective so you can then compare to see if your test was successful but don’t forget to take note of the before results and where you stand so that you can compare them to the results you get from changing something..
Now you need to think about what you are going to change and therefore test. The most common things to change and test are creative, message, channel and offer. And online people tend to test things like different subject headers, calls to action (buy now versus add to basket) and even the colour of the ‘click now’ buttons. Choose what you are going to test based on your objective.
The best practice when carrying out a test is known as AB Test. This simply means that you split your marketing material into two but only change one half of it. This way you can easily analyse the effect of the change. But remember to only change one thing at a time, this is very important. You only ever change one variable at a time as if you change more than one you won’t know which caused the change in performance.
For example, I have just sent an email and changed the subject header and the time that I normally send it. More people open it but which change caused this to happen? Was it the subject header changing or the time that I sent it? Because more than one thing was changed you can only guess now…
Now, here is a question I bet you are dying to know the answer to…how do you pick your audience?
The selection of people you choose should be random, and the best way to randomise your choice is to take a similar data set and audience and randomly split it I wouldn’t recommend doing it by alphabetical order or by another metric, simply split it. Also, you have to compare apples with apples, not apples with oranges and therefore must send your test to the same audience and not two different types of customers i.e. new and returning.
To get the best and most thorough results, testing has to be a constant thing. Some work, and some don’t!
Lastly it is important to review your results. What was the impact of your test? Did it work? If so, then the new changed and tested format becomes your basis for your next test. If it didn’t work or had a negative impact then revert to your previous best performer. This know as the control. Some tests will yield a huge increase, and some only very small increases. However over time incremental small increases can soon add up to be significant so constant testing is key!
If all this sounds a little confusing, don’t worry, we can help you put together a basic test plan and support you until you are comfortable to fly the test nest, plus, there are some great programmes available that you can plug in to your site, Optimizely! is one, check it out.
Next week we will look at gathering customer data and how the cool cats do it! In the meantime please remember to give me your comments.