Test multiple variables allows you to test many variables (variables = each object on the landing page such as logo , button , text , image , another button , another call for action etc.), The test is carried out simultaneously examining all variants or elements on the page, it’s like running many simultaneous A/B tests in a short time and with similar traffic volumes (all the amount of website traffic is divided across all tested pages)
This test has a clear advantage, take for example a headline, image and “calls for action button” you would like to know what title, what image and what call for action button works the best for you.
Variable 1 – “Title”: A or B
Variable 2 – “Image”: A or B
Variable 3 – “call for action”: A or B
(3 page examples would look like this: AAA, AAB, ABA) multivariate allows us to check much more than 3 variables, you can check 10 easy.
Do not forget that there is a small price to pay, we want to create synergy between the collection of objects or variants on a given page, so using this method might be better if you take into account those interactions between objects, it is all a “game” or manipulations of traffic, cost/price (both traffic and time spent), and an complete overview of all elements on a page, I hope I was clear about it , it is not a sexy subject …I know…but it is one of the most important ones , marketing is very emotional , we can look at a beautifully designed page that doesn’t work while the ugly one works like mad , I will not discuss any other methods as they are very complex and requires expertise ,for example methods of behavioral adjustment (based on user behavior on landing page…ok forget it , I will write a post just on this method because it’s an amazing one)
You can perform a test for number of elements on the page
Requires less information than A / B tests
Many small tests (though tests are larger than A / B tests)
Requires knowledge and experience
Requires statistical threshold = takes more time
While you are testing, already in the process try to produce a sufficient amount of landing pages, different versions that you can look over time in order to reach the correct page structure that produces the maximum results, new and different versions that you can test over time, test different concepts, image, hard sale, community and try to find out what works best for you.
you may not want to use all of your traffic for one destination, test your channels, you may want to examine a separate traffic channel (as I mentioned before, traffic from search engines is not similar to traffic from other sources such as a website, social destination etc.) you can define test groups, there is a lot of sense testing different groups of people that behave differently upon reaching the landing page and groups split testing only does good and it effects our conversion ratios.
Install a Tracking software / Analytics
there are many analytic software’s , from Google to omniture , even if you have such a system in place don’t think for a minute that you can retire and leave it running on auto pilot – if you think that you are dead wrong! you will have to follow the numbers, read it like a bible at least 3 times a day, which pages work better, how many actions/sales are performed on a page, where do people exit the page, How much does this “pleasure” cost? etc., the more you are exposed to more data you will understand more about your traffic and you will be able to produce more sales.
Collect the information
Direct your traffic to your leading or best performing landing pages , do so only after enough information is gathered try to be objective and choose your winning page only when you are sure it’s the best performing, do not count on the amount of views or the amount of unique users it doesn’t have much significance, only the conversions matter (not when you have a content website of course) rely solely on your conversion rates, if you do not have enough traffic you might be forced to lengthen the test, accept it with understanding
Analyze your data
Please try to analyze your statistics and try not paying attention to something that happened accidentally or randomly, significant changes can be made only if you look at the full and true picture of things, small changes will not be meaningful if you haven’t gathered data (assuming you have a lot of traffic then you’ll find little changes can result in a huge improvement of hundreds of percent – but only if you have enough traffic!) you know this story for sure , you managed to persuade your VP to change the homepage or landing page , you gather the team , you brain storm , you discuss the new page look & feel, everyone understand the importance of such change…they are all ready…the day comes ,you press the button…AND…BOOM the new page is up and numbers are going down, “my boss will kill me” , “I cant believe it’s not working”…you get these thoughts and you run to your webmaster and tell him change it back man change it all back…STOP , Relax and wait…do you really think that you can measure changes after a few min? wait and gather enough data and then determine whether to change your homepage back to what it was…don’t panic and be mature about it, testing is part of the idea , sometimes it works , sometime it doesn’t and even if it doesn’t – you LEARN from it.
Examine your improvement
You might reach a conclusion and realize that a page is working for you, even if you rely on speculation, whether it’s the trial and error way or even data collection methodology always… always… try to improve, check and examined whether the change is seasonal, temporary or whether it’s continuous, web traffic is influenced by many number of factors, it’s effected from almost everything, even things that happen outside of it, so never pass to auto-pilot, always examined and improve, implement and improve again , it’s an ongoing process …. No rest!