A multivariant video segmentation system and method including accepting video segments that make up a video, accepting video segment variants for the video segments, and selecting a particular video segment variant to provide as a video segment at a particular sequential time order. The video segments are provided to a user at each relative sequential time order via a video player on a first webpage, user inputs associated with the video player and at least one webpage are accepted, and engagements and conversions from the user inputs are determined and stored. The engagements include at least a viewing time of the user viewing the video, and conversions include completion of a predefined task by the user that differs from the user viewing the video. The video segment path with the highest engagement or conversion value associated with information related to the user viewing the video is provided to the user.