Definition
"Traffic Distribution", relates to what percentage of your visitor traffic you are going to put through your experiment.
So, let's say that you allocate 40% of traffic to your experiment and you have the original and one variation more. By default, 20% of visitors will be shown the original, 20% will be shown the 1st variation. This can also be changed for an uneven traffic distribution between variations.
Statistics and goal conversions will be gathered on this 40%.
The rest of the 60% of visitors will be shown the original without collecting any statistics on them.
Change Traffic Distribution
To change the traffic distribution value go to your "Experiment Summary", and adjust the "Traffic distribution" section.
Different Traffic Per Audience
If you want to do different percentage bucketing per audience, you need to create an experiment per audience and allocate the desired traffic to it, via our experiment traffic distribution setting.
Let's say you want to allocate only 20% of US traffic to an experiment, then you need to create an experiment for that US audience, and assign it only 20% on its traffic distribution setting.
What about being able to set different variations to different amounts of traffic?
+1 to Brian's comment, how to you divvy up traffic between variations?
+1
Currently, traffic distribution is only allowed on experience level (not variation level).
I think it is solved now and it is easy to make that
https://i.ibb.co/ZMWrc6N/Screenshot-2.png