Biotraxion
Sign in Request a pilot

Guide

When a step goes wrong: declare the deviation as it happens

Published October 25, 2026

An incubation ran longer than planned. Three weeks later, QA opens the file and reads "incubation extended". By how much? Who saw it? What was done next? Nobody remembers exactly. So someone asks the technician, rereads the notebook, looks for the incubator's timestamps.

This page covers one narrow point: a deviation declared on the step, by the person at the bench, and what the record keeps of it. It does not replace your deviation procedure or your validation.

The reconstructed deviation

When a step goes wrong, it is never a good moment to write. The timer is running, the sample is waiting, you recover first. The deviation gets written later, from memory.

The text that comes out is probably right. But the time, the duration, the order of actions and what was done next are already reconstructed. The longer the delay, the more QA is reading a story.

US GLP asks that unforeseen circumstances be noted when they occur, and that deviations from SOPs be authorized by the study director and documented in the raw data (21 CFR 58.33(c) and 58.81(a)). Those sections say what to document and who authorizes. The rest is your procedure.

A deviation written in the evening is a story. A deviation declared on the step is a record.

What the record keeps

According to the release notes of 18 October 2026, the session report sets, for each step, the written method next to what was observed during the run: times, deviations, photos, pauses and results. Two filters narrow the view to steps with events, or show the method. QA reads the deviation where it happened, next to what was planned.

An incubation overrun follows the same logic. It is displayed after the deadline and recorded as a deviation. The first observation is kept, and the final duration is recorded at validation (release notes of 14 October 2026).

One limit, and it is plain. The record keeps what was observed or declared. It did not see the incident. A deviation nobody declares does not reach the file, whatever the tool.

Nine categories

A declared deviation carries a category: missed step, order, parameter, timing, sterility, reagent, equipment, safety or other. It is shown in the run record and in the report. The record also recovers the step of already sealed deviations.

A category lets you sort. It says nothing about the effect of the deviation on the results. That assessment, and the corrective action, stay with your study director and your QA.

Can you count deviations by category across a whole study? The release notes do not say.

A category helps you find things. The judgment stays human.

The photo that blocks

A step can require a photo. Until it is attached, validation is refused, by click and by gesture. The photo joins the session's results, offline included (release notes of 18 October 2026).

The evidence is there when the step is validated. Nobody chases it afterwards.

Two limits. A photo shows a state: it does not say why, or who decided to carry on. And a hard block raises a practical question: what does the technician do when the photo is impossible?

What the application does not settle

It does not replace your deviation procedure. Who authorizes, within what time, with what consequence for the study: that belongs to your organization. What does the application carry of that circuit (authorization, informing QA)? That is a question to put to any vendor, us included.

It does not settle the clock. The release notes state that the reception time of steps is set by the server (19 October 2026). Which time is attached to a deviation declared offline? Per-step signature, the clock and the chained trail are covered in the guide on e-signature and Part 11.

It does not replace validation of the computerized system, which remains each organization's responsibility.

Six questions for any system

  1. Can the operator declare the deviation on the step, while it is running, without switching tool?
  2. Which time is attached to the declaration, and who sets it?
  3. Does the deviation have a category, and can you count them across a study?
  4. What stops a step from being validated when a required item is missing, and what happens when the item is impossible?
  5. Who must take note of the deviation, and is proof of that in the file?
  6. Does a third party receive the deviations as they were declared, with a way to check they were not altered?

For the last one, our answer is the Audit Pack: protocol, signatures, deviations and hash-chained audit trail, exported in one click as a sealed deliverable your sponsor and your inspector can verify.

See how your deviations are reconstructed today

We show a sample Audit Pack and go through where your deviations are reconstructed after the fact today: contact@biotraxion.com, or request a pilot from the home page. Product changes are dated in the release notes.

Keep reading

  • Modular registry or one tool per function: what an audit gains
  • Signing every step: what an inspector can check
  • Bench execution record vs ELN: what changes at inspection
  • The Audit Pack: what a sponsor can verify on their own
  • Several sponsors, one laboratory
  • ALCOA+ from the bench: what each letter asks of you

← Back to home

© 2026 Biotraxion, activité portée par KIMA HOLDING, SARL au capital de 3 300 €, RCS Toulouse 999 711 682, 5 rue Jules Claretie, 31400 Toulouse.