so you can assign every user, and queue of the variation they would see

but fire an event for 'pop-up viewed' (or similar) - then in GrowthBook, you can add a metric for that event, and attach it to the experiment as an activation metric to filter out all those who were assigned, but never saw the experiment