Eric Metelka

Can you provide more details as to your implementation and the order of operations with regards to segment

Please update the experiment dates in the upper left corner. This will filter to users assigned on or after that start date

If it's sever side, there shouldn't be flicker. Flicker is caused by variant loading client side after the page has begun loading

Microsoft published research on interaction effects a few years ago. My take on it is that mutual exclusion is not needed in most cases https://www.microsoft.com/en-us/research/articles/a-b-interactions-a-call-to-relax/#:~:text=A/B%20interactions%20can%20be,two%20tests%20in%20the%20pair.

The API and SDK are very similar in work to implement and use. The biggest difference is managing caching with the API, depending on how you want to handle.

Are there any specific features that are missing from the options I provided?

Hi Lu - we do not recommend this action. The reason why is because it decreases the amount of traffic to a running experiment. When you do this to a running experiment specifically, you create a traffic imbalance which makes the results unreliable. I'll explain with an example. Imagine you have an experiment exposed to 100% of traffic and now you want to only expose it to 80%. In that 20% being removed, you have a mix of users seeing control and treatment, but it will not be perfectly 50/50 between them. So you're going to be removing more of one group than the other. That means that your remaining sample is now off as well, it's no longer 50/50. The recommendation would be to conclude the current experiment and then take your rollout action.

It is not currently available, but we are working on it. In the meantime, you can use the calculator we provide on the website https://amplitude.com/calculate/sample-size

Hi Oumer - Have you used the duration estimator in Amplitude? Here's a walkthrough of how to use it https://amplitude.com/docs/feature-experiment/workflow/experiment-estimate-duration