The next challenge is to develop changes that would control or eliminate the hazards found - thus improving future performance. The task objective is to first identify any candidate changes that might improve performance, evaluate and rank each, and support the selection of the most effective, efficient performance improvement actions. Use of INVESTIGATION CATALYST for this effort is described in the Recommendation Development Tutorial.

The task is best performed when all the hazards have been found, but may be performed whenever a hazard is identified if the need for action is urgent, as after an accident.

The Analyst examines every Diamond tag on an Event Block or link to determine how it might be changed to improve future action. Thought starters for finding such changes are suggested by the software.

Default criteria are provided for evaluating and ranking each option identified. Users can substitute their own criteria if desired.

The software also provides for the consideration of any tradeoffs that argue in favor of or opposed to implementation of the higher ranked changes, to arrive at selection of the "best" option.

During design stages, the evaluation criteria can be applied to any suggested changes after a hazard has been clearly defined.

During construction and system start-ups, the attributes assigned changes may need to be reviewed if surprises occur, to determine what updates might be necessary.

During ongoing operations, the documented rationale for the evaluations may have to be updated as needed to accommodate the new circumstances.

Whenever an accident investigation discloses new hazards and introduces the need for modification of the current Matrix, the prior analyses of changes should be reviewed

In preparations for shutdowns, users should review the Matrix and Diamonds to determine if shutdown procedures are compatible with prior hazard analyses, and if so the above change development process should be repeated.

The computerized Matrix and Diamonds summary can make these follow-up analyses very efficient, if the matrixes are kept current.