This is exactly the question we hear a billion times every time somebody enquires about our email optimisation service. And the answer is always the same: no idea, just test it.
But unfortunately, for some really strange reason, that’s the point the conversation often ends. Testing sounds extremely complicated when in reality email tests are perhaps the easiest to implement, and the return you would normally get makes not just the initial test but any future tests cost effective. So today I will try to give you a detailed breakdown of email frequency tests, and eliminate any excuses from your vocabulary.
How to run an email frequency test
1. Decide on the variables and the number of them
Firstly you need to create the basic rules of the test. How many different frequency variations do you want to test? We usually run five:
- Twice per month
- 3 times per month
- 4 times per month (Weekly)
- 6 times per month
It is important to mention that this will differ based on the industry you are operating in, and mostly important the life cycle of your product. If you sell £2 million houses when perhaps sending the email six times to the same person would be a bit over the top; in the same way you will be wasting your time testing monthly frequency if you are selling events tickets or ‘hot travel deals’.
2. Develop some appealing offers
Actually this should be done for your ordinary emails anyway, but it’s especially important during the test. With a high frequency schedule you need to make sure you’re delivering highly appealing offers (don’t even think about spamming your respondents with every single new product) because the chance of unsubscribtion or even a spam report will increase with the amount of emails you send. Make sure the offers really are different as well – there is no need to send the same product offer to the same person multiple times (3% or 5% off the same product are not really different in your subscribers eyes).
3. Divide your list
You need to equally divide your mailing or contact list between the number of sending frequency variants (i.e. five sending frequencies then split your list five ways). Make sure to use random segmentation in order to avoid any indirect correlation and achieve the most reliable results afterwards.
4. Decide on the schedule
At this point you should already have several subscriber lists each assigned to a specific sending frequency, along with a set of potential offers you are planning to use. It really helps to visualise the sending schedule so you can simply follow the original plan:
5. Tag the links
Again this should be done in every single campaign you run anyway, and if you don’t really understand what I am talking about then perhaps the frequency test is a bit early for you. Here’s the guide for those who need a reminder. It is impossible to run the test without proper tagging because you simply won’t be able to assign conversions to a specific list and thus conclude the test results.
6. Choosing the winner
Wait at least couple of days after sending the last test offer in order to gather all the results (depending on you industry, quite a solid percent of conversions can come even after 2–3 days after the email is sent). Then combine all offers sent to the same list so you can compare total results. There are two metrics you should be considering when trying to choose the winner:
As you can see the list E (six times per month) has logically generated the most amount of revenue. Sounds great, isn’t it? But what if we do generate maximum revenue but at the same time piss off the majority of our subscribers by sending them offers six times per month? That’s why we need to introduce the secondary set of KPIs in order to make sure that our highest earning frequency at the same time is not harming the list growth. Increase in revenue is always a nice thing in the short term but if we lose all our subscribers, or moreover gain a massive amount of spam complaints, then we are going to lose quite a lot in the long run.
Note that the unsubscription rate for List E is below the average and complaint rate is just a little bigger but still nothing to be worried about. In this case we can confidently conclude that our List E (sending six times per month) is a clear winner, opening an additional opportunity to increase the revenue generated from the regular email campaigns.
Delimitation: We do understand that sending various amount of emails to our test lists will result in several external variables (seasonal effects or extraordinary appeal of some specific offer) so we strongly recommend to monitor the performance of the winning combination for some time after the test in order to ensure that these external variables haven’t affected your initial results.
So what’s next?
Actually the example I’ve used above is the most common result of the frequency test. Surprisingly the majority of businesses do not send emails often enough and are literally wasting potential revenue. I really hope my brief guide will help you to create your own frequency test and start utilising all the potential revenue email is capable of. Stop thinking – go and test.