This aims to present the early implementation of DDI 2.5 in the French initiative of qualitative data bank named beQuali. We will discuss the limits of DDI for documenting qualitative survey when it comes to describe methodological context, materials themselves and their arborescent archivistic ordering. We have fitted DDI following circumvention practices already established and supplemented it with other standards to reach our needs : document all the specificities of our materials, interplay with archiving technologies and digitisation process, data dissemination and harvesting. Indeed none of the existing metadatas formats (DDI, EAD, PREMIS, METS) can handle the granularity of each purpose : qualitative dataset description, archivistic arborescent materials ordering, preservation metadata and compound object aggregation. For simplicity and performance we had to design an in-house metadata format named MetaJSON that acts as a pivot or switchboard between imported or disseminated ones. We are now developing an open source tool that edits and crosswalks this internal format with others. This tool would be designed to help researchers for deposing data and help us to process materials for archiving and dissemination.