Following my last post I promised I will explain about how to get things done…
Choosing the method of how to change/fine tune/optimize
there are two main actions in conversion tuning:
1. Decide what to change to produce other alternatives.
2. Examining the impact of those changes
Choose what to change and what are the alternative is a very important thing, if you won’t change nor examine the impact of these changes on your conversion you lose lots of money!
In the same breath it is important to understand the benefits and limitations of the optimization method you use, there are several factors that determine what method should be used:
1. Test Size
2. How much traffic your site gets and your own confidence in test results
3. variables Interaction of (I’ll explain that in more detail later on)
Size of the test depends on the amount of various and different landing pages (in the professional jargon we call it “recipes”)
For example: If you check 3 different headlines, 4 different selling offers, and 6 different “call for action” buttons you are actually testing72 different pages (3 * 4 * 6) this applies to your standard A/B testing(I will explain more about it soon) As we increase the amount of variables it is quite clear we increase the size of the test.
conversion tuning is based on statistics, as we collect more data will know how to make appropriate changes in order to produce relatively higher conversions, the amount of information usually associated with your personal “confidence level” what do I mean? Would you like to be right 3 out of 4 times? Perhaps 9 out of 10? Maybe 99 out of 100? Once you’ve selected the “confidence level” you can start your test, making changes without sufficient information could be a big mistake which will cost you a lot of money, do not hurry to change things before you have gathered enough information which can statistically conclude that a change must be done here.
How much traffic do you need?
Well…there are 3 main factors you should take into account
1. Test Size – (again….) and this amount of information is collected in order to reach clear conclusions about the required changes, DON’T MIX various testing approaches!
2. Traffic data rate – Traffic Sources …is your traffic from Search Engines? Is it Referring Sites? Is it from other source? traffic source is an important element in the equation, traffic from search engines is completely different traffic source than affiliate sites or domain traffic or direct mail campaign – Please note that! An email Campaign For example, requires a completely different approach rather not use your web based landing page as the email creative being sent to your customers and certainly not for long, the response to such an email will decrease in time it will go down – even if that same landing page works great on the web.
Traffic quantities are supposed to be measured on a daily basis, and the analysis of conversions must be made per traffic channel and NOT an Average overlook! This is why websites with low conversions need large amounts of traffic to reach a certain statistical threshold that will help decide whether to make changes or not.
This will enable you to make true comparisons and changes (traffic is of course relevant to a certain “recipe” / landing page separately).
3. Change / improvement required (Size) – if you have succeeded in reaching the right “recipes” / landing page it will clearly work better when traffic volume changes often making several changes to a landing page that proved to be working will result in a significant improvement / outcome, the result of changes ranging from 1% -5 % (percentage volume of changes in a page) can be significant but it may take longer to reach a decision, these small changes depending on your confidence level (as I mentioned before) the volume change process will vary accordingly.
So what is variable interaction? Very simple – does a change of parameter or an object/variable or sub-page change positively or negatively impacts another variable on the same page, if they have no influence on each other we would say they are independent or are not connected, for example: We want to sell a car and for that we have created a landing page with two different headlines to see what works better, we will also examine 2 different images/creative = we have created 4 different landing pages based on two variables (your standard A/B testing) if you believe that there is no interaction then you probably believe that there is one title that is better than the other regardless of the picture and vice versa … There is a picture/image better than the other – probably this is not the case, each variable is perceived differently –In internet marketing we preach for engagement! We want to interact by using headers, sale messages, calls for action, etc. I think there is almost always interaction between objects which ultimately affects the results.
Now it’s time to dive deeper….I’ll try to describe a little more in depth on the various methods of tuning/optimization (such as A / B Splits, multi variant testing and – behavioral testing and some automated systems on different technologies that can help you sell some more ) all in my next post…