Table of contents
At a glance
How can exploratory testing provide traceable evidence, and where do we need targeted tests?
Exploratory testing combines learning about the system with deciding what to test next. Document the objective, system version, significant observations, findings and resulting decision. A test without a step-by-step script can still generate evidence. Some requirements or boundary cases may need additional targeted tests. Select the combination that answers the outstanding question for the intended use and risk. A test charter is one practical aid, not a universally prescribed legal format.
- Determine the objective and the system prior to the meeting.
- Observation, error hypothesis, and supported conclusion differ.
- Deduce residual checks from the gaps found.
1. Start with an open, yet limited inspection task
Exploratory does not mean clicking on a surface without purpose. A good exercise identifies potential weaknesses and their observable consequences. "Check the application" is too broad. "Investigate whether a discontinued import could generate undetected duplicates," on the other hand, opens several meaningful testing avenues.
The ongoing selection of additional steps is part of the method. A notable intermediate state may trigger a new run. During this process, it is essential to document the specific question currently under investigation. Otherwise, it will be difficult later to distinguish whether a behavior was intentionally evaluated or merely coincidentally observed. For a small team, a limited scope of technical responsibilities is often more effective than an extensive list of general operational actions.
2. Correctly classify the method from a regulatory perspective
The public GAMP 5 description supports critical thinking and appropriate methodologies. GAMP 5 is not a regulatory mandate and does not render the charter presented herein into a binding template. For the assumed human medicinal product GMP application, Annex 11 requires appropriate validation evidence and a risk-based justification. The specific test method must actually address the relevant question.
The FDA-CSA Guidance dated February 3, 2026 explicitly identifies exploratory testing as a potential approach for manufacturing and QMS software in the medical device sector. This recommendation must not be broadly construed as a pharmaceutical regulatory basis. Nevertheless, the underlying methodology is understandable: an unscripted test can be traceable, without predefining every individual click in advance. The applicable framework and selected method must remain independently justified.
3. Write a brief test charter
The following internal work aid provides an example of a sample import function. The function is designed to recognize existing orders during repeated transfers. The team specifically aims to investigate states between technical acceptance and visible confirmation. This charter does not replace a comprehensive evaluation of all import requirements.
The instruction specifies expected limits but allows for additional trials. During the session, for example, break times may be adjusted. Such variations do not need to be fully predefined in advance. The evaluated conditions and relevant observations must subsequently be sufficiently traceable to enable a targeted review of the findings.
| Element | Completed Inspection Task Order | Why it is required |
|---|---|---|
| Objective | Interruptions and restarts must not result in undetected duplicate orders. | Identify the consequences of professional errors |
| Basis | Import stand B, defined test environment, 12 unique order identifiers | Assign a specific basis for observation |
| Exploration | Interruptions before and after visible confirmation; retransmission of transfer | Unforeseen intermediate states are detected |
| Observation | Reconcile inputs, stored orders, rejections and messages | Not only assess screen alerts |
| Boundary | No productive data; no statement regarding all interfaces | Scope limitation |
| Result | Record findings, verified conditions, and targeted re-inspections | Actionable decisions are enabled |
4. Separate observation from explanation
"The application has lost data" is initially an interpretation. The immediate observation might be: Twelve inputs were sent, but only eleven orders are visible in the list. A data record might indeed be missing. Alternatively, a filter could be excluding an order, or the transmission may not have been fully completed. These possibilities require investigation.
Therefore, document the current state, action, timestamp or sequence, and observable outcome initially. Explicitly state the suspected root cause as a hypothesis. Preserve relevant evidence prior to making any further attempts to alter the state. A screenshot may assist, but is not automatically sufficient if critical information is only available in a log or data set.
5. Example: Independent counting in the event of failure
In the following case, twelve different orders are transferred. In the deliberately interrupted first attempt, seven are to be stored completely. Five are still open. In the case of complete repetition, the seven existing identifiers are to be recognized and the five missing ones created. Thereafter, twelve unique orders are expected, not nineteen.
Who counts nineteen orders after resending will find that seven have already been duplicated in the system. The numbers in the table represent predefined target conditions. An actual run may generate different intermediate states; these must then be evaluated against the current requirements. This example does not assume any specific production or batch replication mechanism.
| Step | Input or state | Expected reconciliation | Open evidence question |
|---|---|---|---|
| outset | 12 unique identifiers prepared | 12 Reference entries | Was is the reference list independently verified? |
| Interruption | 7 completely archived, 5 still open | 7 + 5 = 12 | Is this condition actually demonstrable? |
| Repetition | 12 Identification numbers to be re-transferred | 7 identified, 5 new; total 12 | Are duplications and incomplete states recognized? |
| Targeted Residual Testing | Interruption at another time point | Review the result against the same professional guideline. | Does the evidence also apply to the second transfer? |
6. Turn a finding into a targeted validation evidence
An exploratory finding may reveal a new gap in requirements or indicate a violation of an existing rule. Assign it to the intended use. If the team is unable to reproduce the condition, the finding is not automatically resolved. Investigate which information is missing and which further investigation is needed to clarify its significance.
After correction, the critical root cause condition should be specifically investigated. Additionally, adjacent related functions affected should be assessed. This enables the establishment of a clear, traceable relationship between observation, root cause, change, and outcome. Not every finding necessitates a comprehensive individual report; however, the relationship must be preserved and the decision on utilization must be justified.
7. Assess the achieved coverage
A session without identified defects does not prove that all requirements have been met. Document which objectives were evaluated, which conditions actually occurred, and which areas remain open. Distinguish between "passed," "not passed," and "not assessed," to the extent that these terms are appropriately defined in your procedure.
For critical limit values or clear calculation rules, targeted cases are often particularly informative. Exploratory assessments can additionally identify unexpected interactions. These methods are not mutually exclusive. The FDA-CSA Guidance does not strictly assign scripted and unscripted testing to specific risk classes. Select the combination based on knowledge requirements.
8. Checklist for a valid exploratory demonstration
The following summary provides a concise overview to facilitate handover to a colleague who was not present at the meeting. It should enable them to understand the scope of the investigation, key findings, and the rationale for whether further testing is required or not. The quality of the evidence is not determined by the length of a click-through log.
Retain key digital records in controlled association. A verbal reminder of an unusual condition becomes less reliable with each batch release. At the same time, you are not required to document every trivial mouse movement if it has no bearing on the statement.
- Objectives, intended use, and specific basis described.
- Appropriate test data and expected professional limits defined.
- Traceable relevant conditions and state changes.
- Observations of hypotheses kept separately.
- Assessments and their significance are evaluated.
- Targeted residue testing including correction testing specified.
- Achieved coverage and remaining limits documented.
Project experience and collaboration
Prepare an exploratory test with a clear question
With the process and test teams, we define a specific question, such as how an import behaves after interruption. We establish a known starting state, suitable data and expected operational limits. New observations may guide the next steps during the investigation and are recorded so others can follow them. Where a conclusion requires further evidence, we add a targeted, predefined test. The question determines which method is appropriate.
Frequently asked questions about Documenting exploratory testing under GAMP 5
Is exploratory testing uncontrolled trial and error?
Must all test steps be defined in advance?
Does GAMP 5 specify a particular test charter?
The test protocols described in this article are intended as a standalone reference tool and are not a binding GAMP template. GAMP 5 is a sector-wide guidance document and not a regulatory mandate. You may document the testing objective, scope, and results in another appropriate format; the test's traceability and approved procedures remain critical.
Sources: [3]
Does FDA CSA apply immediately to our pharmaceutical GMP system?
The February 2026 Guidance addresses production and QMS software in the medical device sector. It is not a general pharmaceutical regulatory basis. For the GMP system, you must determine the applicable requirements; methodological approaches may be utilized with appropriate justification.
Does a list without identified errors suffice?
Are screenshots mandatory?
Not universal. Select evidence that demonstrates the relevant behavior. A screenshot may confirm a message but cannot verify invisible data processing. Appropriate digital logs or dataset comparisons may provide more credible evidence. Their assignment and availability must align with the audit objective.
What should we do when a fault cannot be reproduced?
Can exploratory testing be performed under high process risk?
The FDA Computer Software Assurance (CSA) Guidance does not generally exclude unscripted testing for processes with high risk. The validation approach must be commensurate with the potential consequences of errors. For specific critical rules, additional targeted or automated testing may be required; the selection of methods must be justified.
Sources: [2]
How is a test turned into a regression test?
Transfer a relevant incident into a sufficiently defined fault condition with an expected response. After corrections and subsequent related changes, this case can be selectively reused. Its validity for the current state must be maintained or re-evaluated.
Sources: [1]
How much subject matter expertise is required of the tester?
He must understand the intended use, typical operation, and potential consequences to effectively evaluate and interpret observations. Technical and subject-matter expertise may be distributed across multiple individuals. This method does not replace qualification; instead, it strategically leverages existing knowledge and clearly identifies its limitations.
Sources and review date
Source date: 7 October 2026. Regulatory requirements, guidelines, and industry practices must be categorized separately. Tables, figures, and case examples are to be considered standalone reference materials; they do not constitute completed product testing.
- EU GMP Annex 11 – Computerised SystemsRevision January 2011; listed in the human medicinal product section. Principle, §§1–17.
- FDA: Computer Software Assurance for Production and Quality Management System SoftwareFinal guidance of 3 February 2026. Non-binding recommendations for production and quality management system software in the medical device sector.
- ISPE: GAMP 5 Guide, 2nd EditionPublic publisher description and edition information; no access to paid full text.
Sources checked: .
Jointly develop the specific evidence process
We help your team define what an exploratory test should investigate and link the findings to targeted test evidence.
- GAMP 5 consulting →
Understand how GAMP 5, EU GMP Annex 11 and 21 CFR Part 11 apply to your systems. We translate the relevant requirements into proportionate validation work and evidence your team can defend in an audit.

