The Traffic share column in the Experiments table on your Overview page (Get Started in the sidebar) shows each experiment's share of all the visitors your experiments have collected. It does not control anything, and it is not a traffic split you can change.
What the column shows
Traffic share is a read-only number. It is an experiment's visitors divided by the visitors of every experiment in the project, including paused and completed ones.
That is why the shares on the Running tab usually add up to less than 100%. The rest belongs to experiments that have finished or are paused.
An experiment that has been running longer usually shows a larger share, because it has had more time to collect visitors.
Why it isn't a 50/50 or 60/40 traffic split
If two experiments show 60% and 40%, that does not mean traffic is being divided 60/40 between them. It also does not mean one experiment is limited to 40% of your visitors.
Both experiments can already be open to 100% of your traffic, and neither one limits the other. Every visitor can enter both, unless you place them in a mutually exclusive group.
If you see an uneven split between Control and Variant inside one experiment, it's usually normal variance while traffic is still low. If the split is still off once the experiment has more traffic, Mida shows a sample ratio mismatch (SRM) warning on its results page so you know something is wrong.
How to actually control traffic
Don't use the Traffic share column to change how traffic is assigned. Use the Traffic Distribution bar on the experiment's Configuration step instead.
The bar sets two things at once:
How much of your site traffic enters the experiment. Anything in the grey Excluded part never sees the experiment. This is the traffic allocation. See What is Traffic Allocation? in General FAQs.
How visitors who enter are split between Control and each variant. For example, a holdout setup of 75% Control and 25% Variant.
Neither setting changes the Traffic share figure on the Overview page.
If you want two experiments to avoid sharing the same visitors, place them in a mutually exclusive group. See How to Create a Mutually Exclusive Group.
An example
Say your project has just two experiments, both open to 100% of traffic and not in a mutually exclusive group.
Test A has been running for 14 days. Test B has been running for 7 days. Traffic share might show about 60% for Test A and 40% for Test B, because Test A has collected more visitors so far.
Every visitor can still enter both experiments. Inside each one, the Control vs Variant split is set separately on its Configuration step.


