Migrating tens of thousands of files from SFTP into Bynder
Moving an unstructured media library into Bynder: taxonomy rebuilt from the folder tree, traceable deduplication and an indexing acceptance review.
The engagement
Campaigns, e-commerce packshots, lookbooks, videos and InDesign/Illustrator source files lived on a vendor SFTP server: tens of thousands of files, close to 500 GB, with no structured metadata whatsoever. The target Bynder DAM was empty at kickoff and its taxonomy had several dozen metaproperties to populate. The only usable signal for rebuilding indexing was the file paths and names.
We mapped the source in full, then replicated the SFTP with rclone, checking integrity file by file. A rules engine written in Python and PowerShell then derives Bynder metaproperties from each file's path, name and extension. Metaproperties that cannot be derived from the folder tree were documented separately: they require contributor input or a product mapping. The scope was partitioned file by file (e-commerce packshots, campaign content, creative source files, technical artefacts excluded), and InDesign packages were audited then reassembled with their links and fonts.
Deduplication was handled by archiving, never by deletion, with a line-by-line audit trail of every discarded copy. The portal is now populated and indexed, with no duplicates and no missing files recorded at acceptance, and coverage of the structuring fields moved from marginal to near-complete. Control collections and a remediation workbook drive the remaining metadata work.