The subject technology receives, from a document object model, a set of nodes of a web page, each node corresponding to an element of the web page. The subject technology determines, using an intersection observer, a position and visibility of a set of first elements within a viewport of the web page. The subject technology determines, using a mutation observer, a set of new elements within the viewport that was previously hidden and now visible. The subject technology provides a set of visible elements, based at least in part on the set of new elements and the set of first element, as elements that have been seen by a user during a session. The subject technology generates an index based at least in part on the set of visible elements, the index comprising information corresponding to text that was visible to the user during the session.