I don't think we record that information right now and it can effect performance perception quite a bit to have the late-loading banner.
Ideally we wouldn't just track that the user was targeted by a banner, but also the time when it was displayed to the user.
This will be particularly useful in the context of T187299: User-perceived page load performance study where we will be able to verify if the late appearance of a banner pushing down content negatively affects user sentiment about performance.