LIMS selection: a sample workflow checklist

A sample-tracking workflow, an experimental notebook and a store of instrument files are different jobs. Suppliers may combine LIMS (laboratory information management), ELN (electronic lab notebook) and SDMS (scientific data management) in one platform. Write down which job you need and which system owns each record. Product labels alone do not settle scope. This sample-centred trial does not cover a clinical patient-order-report workflow, diagnostic interpretation or every quality-control process. Ask the responsible laboratory and quality owners to define those separate requirements.

Use invented identifiers only. Create parent sample SYN-LAB-001 with 10 mL and two aliquots, SYN-LAB-001-A and SYN-LAB-001-B, each 4 mL. Expect a remaining 2 mL if your agreed process tracks volume without loss. Then attempt a third 4 mL aliquot: the requested total is 12 mL. Define in advance whether this must be rejected or raised as an explicit exception; a silently negative balance must not look like a successful split. Follow each child back to its parent and each parent to its children. If your workflow does not use aliquots or volume, document that scoped exclusion and replace it with your real relationship test.

Define a numeric test parameter, Synthetic Analyte A, with the expected unit mg/L. Prepare a synthetic file with SYN-LAB-001-A at 5 mg/L, an unknown sample ID SYN-LAB-999, and a third row at 5 g/L. Agree permitted conversions first: 5 g/L means 5000 mg/L, never 5 mg/L. An unconfigured unit must follow the agreed rejection or review path. Observe whether the whole file is rejected or valid rows are committed while invalid rows are held. Correct the file and establish the valid 5 mg/L result before testing its correction. Retry an unchanged valid file and separately recover a partial import; neither should create an unnoticed duplicate. Change a column heading during the trial and identify who can inspect and maintain the mapping. Record any custom script, connector or paid configuration required. Structural import checks do not test all scientific limits, out-of-specification handling or instrument QC.

With separate operator and reviewer roles, correct the example value from 5 to 6 mg/L using the reason synthetic correction. Find the old value, new value, actor, time and reason together with the original raw file. Try changing an approved record and replacing its attachment. Check the agreed reopening or revision process and whether a reviewer must approve again. This is an observation of the tested settings, not proof of an immutable audit trail or a legal electronic signature. Repeat affected cases after changing the version, configuration, integration or permissions.

Export the sample, aliquots, results, attachments and relevant history in the offered formats. Have a second person reconstruct the parent-child links, original raw file, corrected result and reason using the export and its documentation. A flat table of final values may not preserve those relationships. Check stable identifiers, timestamps and units; record any missing element as open. Check that attachments are supplied as actual readable files, not only internal links requiring the original session. Record how a missing or unreadable attachment would be obtained. This read-back tests portability, not a full disaster-recovery plan or proof that another product can import the files.

Start each decision as Open / untested. Record Pass only for the agreed expectation you observed, Fail for a demonstrated mismatch, and Open / untested for unavailable access, missing evidence or an unrun case. Keep the reason and next action. Name mandatory requirements before the trial; a failed or open mandatory case prevents a positive readiness decision for that scope. Do not average it away with other successes. A justified exclusion needs an owner and a reason; it is not a pass. This plan provides no overall score and does not change the Transparency Score or provider order.

Keep the exact modules, hosting option, users, instruments, interfaces and migration volume with the quote. Separate a vendor claim from your own observation and from a remaining question. Attach the trial worksheet to your procurement brief, assign unresolved items and agree what must be demonstrated next. Use synthetic data throughout; do not upload patient records or proprietary research data to a sales trial without the organisation’s separate approval and safeguards. This is not system validation, a clinical or regulatory approval, or a recommendation to replace an existing laboratory system.