Your inventory says the tube is in C4. The freezer says otherwise.

The tube the record says is in position C4

A scientist needs aliquot ALQ-1188 for a repeat assay. The inventory says freezer FZ-3, box B-12, position C4. Position C4 is empty. The tube was used up in March, and the only record of that is a line in someone’s notebook.

The same week, a CRO returns shipment SH-031. Its manifest lists CR-77-014 to CR-77-040. Nobody can say which internal aliquots those were, because the outgoing list lived in a spreadsheet that has since been edited. And when an outlier result appears, nobody can trace the aliquot back to source lot SRC-0419, because the split was recorded as a comment rather than a link.

A record that no longer matches the material is worse than no record. People trust it until it fails them.

ALQ-1188, SH-031 and the other identifiers in this guide are illustrative, not client data.

Follow one sample type through every move

Pick one sample type and one repeated movement, such as aliquots sent to a CRO and returned. Walk it with the people who handle it. Note where each event is written down: registration in the ELN, location in the inventory system, consumption in a notebook, dispatch in a spreadsheet, receipt in a courier email.

Then ask the hard questions. Who records a thaw? Who notices when a tube is used up? What happens to a label that will not scan in the cold room? The answers show where the record and the freezer drift apart.

Give every physical item one identity and a parent

The NCI Best Practices for Biospecimen Resources give a useful baseline even outside biobanking. Each specimen and each derived sample gets its own unique, system-generated identifier. Labels are both machine- and human-readable. The origin of every aliquot is recorded, and the identifier never encodes the storage position.

That last rule matters more than it looks. An ID like FZ3-B12-C4 becomes wrong the first time the tube moves. Keep location as data about the item, not part of its name.

• Split: ALQ-1188 is a child of SRC-0419, with the volume taken and the date. Results tie back through that link.
• Pool: a pooled sample lists every parent, so one contaminated source can be traced to everything it touched.
• Relabel: a damaged label gets a replacement, and the old barcode stays recorded as an alias. A scan of either finds one item.
• External IDs: CR-77-014 is stored beside ALQ-1188, scoped to that partner and shipment, never as a replacement.

Record state changes when they happen

Agree a small set of states with the lab: registered, stored, checked out, consumed, shipped, received and discarded. Then agree which transitions are allowed. A consumed tube cannot be checked out. A shipped tube cannot be stored until it is received.

NCI recommends updating the database in real time each time a sample moves, including its position in the box, rack and freezer. In practice that means a scan at the bench or freezer, not a typing session at five o’clock. It is also how records stay contemporaneous, one of the ALCOA properties in FDA’s data integrity guidance.

Track partial use too. NCI includes the amount distributed and the amount remaining. A tube with 20 µL left is not the same as a full one, and a request that needs 50 µL should know before someone opens the freezer.

Reconcile every shipment on both sides

NCI describes a shipping manifest sent ahead to the recipient, a copy inside the package and a check of every label against the packing list on arrival. Discrepancies are documented and reported to the sender. That is the contract to build around.

When SH-031 comes back, each returned tube is matched to the item that left, not created as a new record. An unknown ID waits for review. A partial return keeps the shipment open, so nothing is quietly lost between two buildings.

Plan for mis-scans, duplicates and missing tubes

Real labs produce bad scans. Labels frost over, a tube is scanned twice, a box goes back on the wrong shelf. Design each case with the lab before writing code, then test it:

• Scan the same tube twice in one move. One event is recorded, not two.
• Scan an unreadable or unknown label. Nothing changes until someone resolves it.
• Check out a tube already marked consumed. The scan is blocked and the conflict is shown.
• Return a shipment with one tube missing and one extra. Both appear as open exceptions.
• Correct a wrong move. The correction is a new event and the original stays in history.
• Sign in as a user from another project. Its samples, positions and shipments stay hidden.

Run a periodic physical audit as well. Scan one box at a time and compare it with what the record expects. Scheduled checks catch the events nobody scanned. NCI also recommends role- and project-based access control and a tamper-proof audit trail that records who changed what, when and why.

Check what your inventory already does, then build the gap

Most labs already own part of this. Benchling’s inventory data model documents barcodes, box row and column positions, an archive reason and a checkout status of available, reserved or checked out. Its transfer API moves a quantity from a source container into a destination container. Other LIMS and inventory products offer similar features. Configure those first.

The gap, when there is one, sits at the boundaries: scanning that fits the cold room, parent links that survive a split, shipment reconciliation with a partner who uses their own IDs, and an exception queue someone actually owns. We build that boundary layer and hand it over with tests.

Scope one sample movement you can trust

Request a call from this article and describe the last tube or shipment your team could not account for. The first conversation needs no data; confidential files can wait for an NDA or your approved access process. See how we work with biotech teams. If results from those samples come back from a partner, read the CRO data integration guide; for requests between internal systems, see the ELN and LIMS integration guide.