We should ideally only reindex if there would be a different index in the end.  We already have some mechanism for that but they can be improved with more context sensitive data anyway. This data is mostly already saved I  the data base but it should be reduced like described in an other user story.