Little advice for split testing (11)
Hi guys,
I read in a guide that when I have multiple landing pages in order to do good split testing (in my tracker) I should create a single campaign and the let the tracker split-test the landing pages evenly. But I suppose that in this case this shouldn't be done because I have created different banners of different colors and thought was good thing to make landers follow the banner design style.
What do you suggest? One campaign with different landing pages or multiple landing pages so that "x color banner" leads to "x color lander"
If you want to test, whether using the same style/colors from the banner to the LP makes any sense, then no, do not rotate anything and send the users from a "blue" banner to a "blue" lander.
From my experience, this doesn't really work, it's something I tested long time ago and didn't really see a strong pattern to confirm this theory. Sometimes it worked, sometimes it didn't. I'd say color is one of the weak factors to optimize.
Thank you Matuloo 
. There are so many guides available all around and some teach different things . Sometimes it is said to split-test lander evenly and sometimes that we must follow "the same color" rule and that can be confusing
Originally Posted by davidep
. There are so many guides available all around and some teach different things . Sometimes it is said to split-test lander evenly and sometimes that we must follow "the same color" rule and that can be confusing In your case, you want to test several colors and whether it has any impact on the performance of the funnel.
In theory, if you really want to make a PROPER test, you would really have to take all the landers and all the banners and run a split test with all of them. Then you would have to drill down in the data and check whether the blue/blue combinations converted the best or whether the color combinations didn't matter.
But with even 5 colors, you would be looking at a mad number of combinations that would cost a lot of $ to properly test.
In this case, it's better to run blue banner to blue lander directly and then compare the results with a "mixed" campaign that you probably already had running before.
Mat.
Testing unified colours and a well flowing funnel is always a good thing to test though it wouldn't always work and is probably absolutely unnecessary if you're buying traffic from mDSPs with thousands of placements, especially when talking about pure AM campaigns. Getting mixed results is definitely not going to be a surprise here.
Banner to lander to offer compatibility would largely depend on the (one)placement or direct buy you're testing, the traffic type and whether you want your funnel to be more recognizable (i.e. brand), trustworthy and so on in case you want to catch those conversions.
In other words - I wouldn't waste too much money testing this sort of stuff with pure AM campaigns that have chaotic traffic coming in from every direction combined with somewhat shady offers and not very unique landers.
Vincent James quote "test the things that scream" the offer, the angle, the price... Stuff like color will need so much traffic to reach statistical significance.
Thank you guys, you all are giving me precious advices .
@shishev. So you are saying that in case of Mobile Traffic with DSP (which is what I am currently doing) I shouldn't care much about this color thing because of the large amount of placements? Correct?
@rocketstuffideas. Yes things like angles and offers tend to be the determinant part of the game
Originally Posted by davidep
.@shishev. So you are saying that in case of Mobile Traffic with DSP (which is what I am currently doing) I shouldn't care much about this color thing because of the large amount of placements? Correct?
@rocketstuffideas. Yes things like angles and offers tend to be the determinant part of the game
Testing color is a waste of time and money if you haven't already tested the most important parts of the landing page.
Test the headline and call to action first, the main image and text then the entire feel of the lander and find the winners(statistically significant winner, don't just stop the experiment because you think this one is better). Once you get that part optimized, then proceed to other elements with less impact in conversions like the color.
I know there are many case studies that the red button color increased the conversions by 20%, but I can tell you from my own and my clients experience, it doesn't really work like this, you need to have huge traffic and other elements optimized first in order to see relevant changes just by changing colors.
If you have lots of traffic you can test more variations or even try multi-variate, but unless you have, stick with 3-5 variations per experiment, otherwise it will take too much time to reach statistical significance and in aff marketing you don't have a lot of time.
Hey guys, I've a question to ask hoping it's not an idiot one :P. I was wondering whether for some offers it is better to run them in the night time rather than in the day time. Did it ever happen to you that some campaigns perform better because of the period of the day they've been run?
Originally Posted by davidep
Home > Paid Traffic Sources > Mobile