We create a new collection on each deploy, so that in case the structure of the data changes, it's automatically handled on the next reindex