Blog · By IVF Automation team · 17 Oct 2026

IVF lab witnessing and sample identification: a practical workflow

Every IVF lab handles material from many patients on the same day, often under time pressure. A single identification slip at the wrong moment cannot always be undone. That is why witnessing, the independent check that the right sample belongs to the right patient at each critical step, sits at the heart of laboratory quality systems. This article explains what the guidance expects, where identification errors tend to creep in, how to design a witnessing workflow and log, and where barcode, RFID and AI tools fit. It is operational only. Your witnessing protocol, its critical points and its fallback procedures must be approved by your lab director and written into your own SOPs, not copied from this article.

What the guidance and the law say

ESHRE's Revised guidelines for good practice in IVF laboratories (2015) devote a full section to identification and traceability. Among its points:

ESHRE's updated Recommendations on Good Practice in the IVF laboratory (2026) keep these principles and add that traceability data should be captured and validated electronically, with any manual transcription double-checked.

The UK regulator's HFEA Code of Practice (guidance note 18) is one of the most detailed public references on witnessing. In its letter explaining the revised guidance, the HFEA highlights several points that are useful anywhere: label every sample with the full name and a further identifier, do not rely on labelled lids because lids move, witness irreversible steps before they happen, risk-assess and monitor any change to the protocol, and give staff refresher training.

In India, the ART (Regulation) Act 2021 states in section 24 that a woman shall not be treated with gametes or embryos derived from more than one man or woman in one treatment cycle, and that a clinic shall never mix semen from two individuals. Section 23 requires detailed records of donor gametes and embryos, kept for at least ten years and open to inspection. Rule 23.2 of the ART (Regulation) Rules 2022 requires case records, consent forms and laboratory results to be preserved for ten years from completion of the procedure. A reliable witnessing record is how a lab shows, years later, that these rules were followed.

Where identification errors creep in

A widely cited failure mode and effects analysis by Rienzi and colleagues (2015) mapped the witnessing process in a busy IVF centre. The team identified seven process phases, 19 steps and 32 possible failure modes. The most vulnerable steps were the first identification at gamete collection and identification at thawing. The causes they named are familiar to any lab manager:

The authors also describe involuntary automaticity: a witness who has confirmed hundreds of matching labels can start to "see" a match without really reading it.

The hidden cost of manual double-witnessing

Manual double-witnessing is still the baseline in many labs, and it works when it is done properly. But a 2021 study in F&S Reports points out its weaknesses: two checkers do not always behave independently, discrepancies can be missed under cognitive load, and responsibility can become ambiguous. Every time an embryologist stops their own task to witness for a colleague, that interruption is itself a risk. In that study's laboratory network, measured witnessing time per procedure, including the interruption to the second person, fell by 58 to 92 seconds with an electronic system, depending on the procedure.

The practical lesson is not "manual is bad". It is that witnessing has to be designed into the workflow, with enough staff, protected time and clear accountability, rather than bolted on.

Step 1: map your critical points

List every point at which material is identified, moved, combined or made irreversible. A typical map includes:

For each point, write down who performs it, who witnesses, what is checked and where the record goes. Mark which steps cannot be reversed. These must be witnessed before they happen.

Step 2: make labels do the work

Step 3: one patient, one workspace

Only one patient's material should be open in a workstation at a time, as ESHRE advises. Clear the space between patients and keep a visible "in use" marker on each workstation. Plan the day's schedule so that two procedures needing the same workstation are not booked back to back without time to clear it.

Step 4: record each witness step properly

A witness entry is only useful if someone can rely on it years later. For each step, record:

Never let a witness sign for a step they did not see. If a check was missed, record that fact and follow your incident procedure. Do not back-fill. Our free SOP templates can help you set out the procedure, and our guide to daily QC logs uses the same record-keeping principles.

Step 5: train, audit and review

Where AI and automation fit

Automation can take some of the burden of witnessing off people, but it does not remove the need for a protocol.

Electronic witnessing systems

Electronic witnessing systems use barcodes or RFID tags on every dish and tube, with readers at each workstation. Such systems typically alert staff when material from two different patients appears in the same work area, and log every step automatically. In the Rienzi study, after a five-month period with an RFID-based system, the centre reported a primary human error rate of 0.05% (4 of 8,628 witnessing steps), with every one of those errors intercepted by the system before a mismatch. An earlier conference report in Fertility and Sterility (2011) is a useful reality check. Some of the overrides it recorded were linked to power cuts, server failures, flow-chart configuration errors and broken tags. Keep a tested manual fallback and validate any system on site, running it alongside manual witnessing before you rely on it. These figures come from individual centres and do not guarantee results elsewhere.

Lower-cost digital steps

Not every lab is ready for a full system. Lower-cost steps include:

Using AI on your witness data

Once witness records are digital, AI-assisted analysis can help managers spot patterns that are hard to see by eye. Examples include clusters of near misses at particular times of day, workstations where checks are often late, or schedules that regularly create interruptions. Use these tools as decision support for management and audit only. The identification check itself must stay with trained staff or a validated system.

When choosing a system, ask vendors about validation support, offline and fallback modes, data export, audit trails and service response times. Our guide to equipment vendor questions covers service contracts, and the vendor directory lists suppliers. Listings are not endorsements.

A note on data protection

Witness logs, labels and scan records contain personal data about patients and staff. Limit who can see them, secure any devices and cloud systems, and include witnessing systems in your data-protection planning. Our free DPDPA readiness assessment helps you check where your clinic stands.

Quick checklist

Sources

Your next step

Draw your lab's critical-point map this week, then compare it with your current witnessing SOP. Our free SOP templates and ART Act checklist will help you put it into writing. To get practical lab-ops guides like this every week, subscribe to IVF Automation Weekly using the form below.

For inspection readiness, see our ART Act compliance checklist. For staffing, use the jobs board, and for more workflow ideas see the workflow library. This is operational information only; see our disclaimer.

General operational information only; not medical, clinical or legal advice. See our disclaimer.

Get IVF Automation Weekly

Compliance, automation and opportunities for India's IVF labs and clinics, every week in five minutes. Free, unsubscribe anytime.

Also: open IVF lab jobs · vendor directory · featured listings