AI GxP validation for regulated life sciences
CSA

Unscripted Testing Under FDA CSA: Charter Examples, Evidence & What Inspectors Expect

What unscripted and exploratory testing mean under FDA's finalised Computer Software Assurance (CSA) guidance, copy-paste test charter examples, what evidence an inspector expects when there is no script, and how Cybroscape GxP Copilot automates it.

Updated Cybroscape Technologies6 min read

Key takeaway

What unscripted and exploratory testing mean under FDA's finalised Computer Software Assurance (CSA) guidance, copy-paste test charter examples, what evidence an inspector expects when there is no script, and how Cybroscape GxP Copilot automates it.

Unscripted testing is the part of CSA that makes quality teams uneasy, usually because of a misunderstanding: people hear "unscripted" and think "undocumented". It isn't. It means you didn't write the steps in advance — not that nobody recorded what happened.

Once that's clear, the question becomes practical: what evidence do you keep, and how do you satisfy an inspector who asks to see the test?

The three modes, plainly

  • Scripted. Steps and expected results written and approved before execution, results recorded against them. Reproducible by anyone. The right choice for high-risk functions and anything you will need to re-run identically.
  • Unscripted with a charter. You define an objective and boundaries in advance — "confirm the approval workflow behaves correctly for a user without approver rights" — then a qualified tester explores. Steps are recorded as they happen.
  • Ad hoc / error-guessing. An experienced person deliberately tries to break something based on judgement. Useful, lightweight, and only appropriate for low-risk areas.

CSA doesn't replace scripted testing. It stops scripted testing being the default for everything. Which mode applies to which function comes out of your risk assessment, not out of preference.

What evidence to keep when there is no script

This is the whole question, and the answer is fairly short. For each unscripted session, record:

  • Who tested, and that they were qualified to.
  • What the objective was — the charter, written before the session.
  • The system and version tested, and the environment.
  • What was actually done — a factual trail of the path taken, not a reconstruction. Screenshots, exported logs, or notes taken during the session.
  • What was found, including anything that worked as expected, not only defects.
  • The conclusion and who accepted it.

A good test: could a competent colleague read this a year from now and understand what was covered and what the tester concluded? If yes, it is evidence. If it reads "tested OK", it is not.

Where scripted testing is still the answer

  • Functions where an undetected error reaches product or patient.
  • Calculations and algorithms producing values used in decisions — you need exact expected results.
  • Electronic signature and access control behaviour, where you are demonstrating a specific regulatory requirement.
  • Anything you will need to re-run identically after a change, which is most of your regression set.
  • Data integrity controls — audit trail capture, record immutability.

Note the pattern: script where you need reproducibility or an exact expected result. Explore where you need to find the unexpected. They answer different questions, which is why mature teams use both on the same system.

Copy-paste CSA unscripted test charter example (and how GxP Copilot generates it)

A review-ready unscripted test session record does not need 30 pages of pre-written screenshots. Here is the structure Cybroscape GxP Copilot generates automatically for requirements scored as Not-High-Process-Risk under FDA's finalised CSA guidance:

  • Requirement ID & Risk Basis: URS-018 (Role-based dashboard filter & export); assessed Not-High-Process-Risk (no direct impact on batch release, safety record or Part 11 signature integrity; high detectability).
  • Test Mode & Charter Objective: Unscripted Exploratory Charter. Verify that a read-only QA Reviewer role can filter records by date and status, export a CSV summary, and cannot modify or approve any record via UI or URL manipulation.
  • Time-Box & Environment: 30-minute session in Validated Staging (v3.2.1) by qualified validation engineer.
  • Contemporaneous Session Notes & Findings: Tested 4 filter combinations, verified CSV row count matches screen table, attempted direct POST/edit route as read-only user (403 Forbidden logged in audit trail). 0 defects found.
  • Disposition & Part 11 Sign-Off: Pass; electronically signed with re-authentication by Tester and QA Approver, automatically linked to URS-018 in the live Requirement Traceability Matrix.

Common mistakes

Writing the charter afterwards. If the objective is reconstructed after the session, the record is not contemporaneous, and that is a data integrity problem rather than a testing one.

Letting anyone do it. Unscripted testing depends entirely on the tester's competence — they are making judgement calls in real time. Record why the person was qualified.

Only recording failures. If the session notes contain nothing but defects, there is no evidence of coverage.

Using it because scripting is tedious. That is the wrong reason, and it shows. The decision belongs to the risk assessment.

For how this fits the wider move, see moving from CSV to CSA and CSA services.

Where to go next

Explore Cybroscape GxP Copilot for AI-native validation, Cybroscape TraceDraft for source-traceable clinical documentation, or book a demo to see either on your own data.

Part of our guide to GxP software — categories, evaluation criteria and how the platforms compare. For the detail on this specifically, see CSV to CSA.

High riskFailure could affect patient safety or product quality
Scripted testingWritten steps, expected results, an independent reviewer
Step-by-step results, screenshots and signatures
Medium riskFailure is noticed and contained
Unscripted testingExploratory or ad-hoc testing against the intended use
A record of what was tested, by whom, and the result
Low riskStandard function, supplier-tested
Supplier evidenceThe vendor’s own testing, assessed and accepted
The assessment and the reason it is enough
Computer Software Assurance in one picture: test effort follows the risk of each requirement. Based on FDA’s Computer Software Assurance guidance.
See it on your own systemBring a URS or a system. We run GxP Copilot against it live on the call.

Sources

Primary documents referred to on this page, from their official publishers.

unscripted testingunscripted testing csacsa unscripted testing examplesgxp validationexploratory testing gxpcsa testing evidencead hoc testing validationcsa test documentation
FAQ

Frequently Asked Questions

Prefer to talk first? Chat with us on WhatsApp or book a call with an expert.

What is unscripted testing under CSA?

Testing where the steps were not written in advance — not testing that goes unrecorded. A qualified tester works to an objective defined beforehand (a charter) and records what they did as they do it. Ad hoc or error-guessing testing is a lighter form, appropriate only for low-risk areas.

What evidence do you keep for unscripted testing?

Who tested and why they were qualified; the objective, written before the session; the system and version; a factual trail of what was actually done; what was found, including what worked; and the conclusion with who accepted it. The test is whether a competent colleague could read it a year later and understand the coverage.

When should you still use scripted testing?

Where an undetected error could reach product or patient; for calculations needing exact expected results; for electronic signature and access control behaviour; for anything you must re-run identically after a change; and for data integrity controls such as audit trail capture and record immutability.

Can unscripted testing fail an inspection?

The method is accepted; undocumented testing is not. The usual failures are writing the charter after the session, letting unqualified people test, recording only defects so there is no evidence of coverage, and choosing the method because scripting is tedious rather than because the risk assessment called for it.

Who decides which testing method to use?

The risk assessment, not preference. Each function's risk rating determines whether it is scripted, explored against a charter, or covered by ad hoc testing — and that mapping should be stated in the validation plan so the choice is defensible later.

Keep reading

All articles
CSADefending CSA in an Inspection: The Questions and the EvidenceThe fear that stops teams adopting CSA is the audit. What inspectors actually ask when you have tested less, the evidence that answers them, and the record that makes a risk-based decision defensible. · 5 min readRegulatoryDo You Need to Rename AI in Your SOPs and Validation Documents?The short answer is no. The longer answer covers where the new terminology will reach you anyway, how to write documents that survive a vocabulary change, and why a global find-and-replace is the wrong reflex. · 5 min readBest PracticesOutsourcing Validation: What to Keep In-House and What to Hand OverValidation capacity is hard to hire and uneven in demand, so most regulated companies outsource some of it. What transfers safely to a partner, what never should, and how to run the relationship so accountability stays where the regulator expects it. · 5 min readBest PracticesValidation Metrics That Actually Mean SomethingMost validation dashboards measure activity, not quality: documents produced, tests executed, days elapsed. The measures that tell you whether validation is working, and the ones that quietly reward the wrong behaviour. · 5 min readCSAHow to Move From CSV to CSA Without Restarting Your Quality SystemA practical transition from scripted CSV to Computer Software Assurance: what changes in your SOPs, what to pilot first, how to bring QA with you, and the mistakes that make CSA look like cutting corners. · 6 min readCSARisk-based testing under CSA: how to actually reduce scripted-test effortThe per-requirement risk model that lets you defensibly cut scripted testing on non-critical requirements. · 6 min readValidationDecommissioning a GxP System: Retiring It Without Losing the RecordsSwitching a validated system off is a controlled activity, not an IT task. What has to happen before shutdown, how to keep records retrievable for their full retention period, and why half-retired systems are an inspection risk. · 5 min readCSAComputer Software Assurance (CSA): The FDA Guidance ExplainedWhat the FDA's final CSA guidance (February 2026) says: who it applies to, the six-step risk framework, high process risk, scripted and unscripted testing, vendor evidence, the record it expects, and how CSA relates to CSV and Part 11. · 12 min read

Bring a system. We'll show you the package.

Book a demo, or ask a validation lead a quick question on WhatsApp.

  • We reply within one business day
  • We run your URS live on the call
  • You get a clear plan for your system
Chat on WhatsApp

Book a demo

Four fields. The right specialist replies within one business day.

Your details stay private. We never share them.