Consider a TAS that adopts a keyword-driven framework. The SUT is a web application and the set of available keywords for writing the automated tests consists of a very large number of keywords that relate to highly specific user actions tied directly to the GUI of the SUT. The automated tests written with such set of keywords are statically analyzed by a custom tool that shows the presence of many repeated instances of several identical sequences of keywords. The waiting mechanism implemented by the TAS for a webpage load is based on a synchronous sampling within a given timeout: the TAS allows checking a webpage load every X seconds until a timeout value.

Based only on the given information, which of the following recommendations would you provide for improving the TAS (assuming it is possible to perform all of them)?