URS (User Requirements Specification) for Regulated Manufacturing Systems
Define what a manufacturing system needs to do, how users will accept it, and how changes stay connected to approved evidence.

On this page · 12 sections
- 1What a URS is—and how it differs from FS, CS and CSV
- 2Where user requirements sit in the V-model
- 3What to include in a usable manufacturing URS
- 4Writing a testable requirement: a weighing example
- 5Linking requirement revisions to design and acceptance evidence
- 6US regulatory baseline: obligations, scope and specification practice
- 7Data integrity: specify record behaviour, not just principles
- 8Interfaces and exception recovery: a failed-transfer example
- 9Who reviews, approves and owns unresolved requirements
- 10Changing an approved URS: a calculation-rule example
- 11Common URS failures and the controls that expose them
- 12Implementation checklist: from approved requirements to CSV evidence
How does URS apply to your shop floor?
Pick your industry and scale — Ask V5 rewrites the definition in your context, gives a worked example, and shows what V5 does on day one.
01What a URS is—and how it differs from FS, CS and CSV
URS (User Requirements Specification) is a controlled description of what users need a manufacturing system to do, the conditions in which it will operate, and the criteria for accepting it.
It describes intended use: the manufacturing activities supported, the roles performing them, and the process boundaries, including interfaces and activities outside scope. Acceptance outcomes describe observable results that let reviewers decide whether the system meets those needs.
| Term | Main question | Typical content | Relationship to acceptance |
|---|---|---|---|
| URS | What do users need? | Intended use, desired outcomes, operating constraints and user roles | Defines the outcomes and criteria used to judge suitability |
| Functional specification (FS) | How will the system behave? | Functions, workflows, calculations and exception responses | Describes functional behaviour to test against user needs |
| Configuration specification (CS) | Which settings will implement that behaviour? | Selected parameters, permissions, options and mappings | Identifies the configured state to check |
| Computer system validation (CSV) | Is the system fit for intended use? | Lifecycle planning, risk assessment, specifications, testing and review | Builds and maintains evidence supporting an acceptance decision |
A user need is not the supplier’s proposed implementation. For example, preventing an operator from proceeding with an unidentified material is an outcome; choosing a scanner and configuring a blocking rule are implementation decisions. Keeping them separate allows alternative designs to be evaluated without silently changing the need.
These boundaries are requirements-engineering conventions, not universal legal document definitions. A URS is a practical way to organise needs supporting applicable obligations. The functional specification and configuration specification develop the implementation detail; the wider validation lifecycle connects those documents to evidence and accountable acceptance decisions.
02Where user requirements sit in the V-model
The V-model is an industry-practice way to connect user needs with design decisions and acceptance evidence, not a prescribed project methodology. Requirements guide what gets built or configured; evidence returns to those requirements to demonstrate whether the intended use is supported.
A practical lifecycle workflow is:
Discovery → requirements review → supplier or design assessment → baseline approval → implementation and verification → operational handover → subsequent change and reassessment.
Discovery identifies actual tasks, constraints and failure consequences. Review challenges ambiguous needs; supplier assessment and internal design review examine whether a proposed solution can meet them. Baseline approval establishes the agreed reference for implementation and testing. Verification produces evidence against that reference, while handover transfers operating instructions, support responsibilities and unresolved issues to accountable owners.
Supplier demonstrations show proposed capabilities; internal design reviews examine the solution’s logic. Neither substitutes for site-specific acceptance checks using the intended configuration, workflows and interfaces. Design verification addresses the related question of whether design outputs satisfy their inputs.
Iteration can occur before and after baselining. Controlled iteration records the reason, review decision and affected evidence; silently changing acceptance criteria to fit a result does not. Requirement-to-evidence traceability connects these lifecycle stages, including later changes, without forcing every project into a single-pass sequence.
- Define intended useIdentify process boundaries, users, operating conditions and exclusions.
- Capture user needsRecord outcomes, dependencies, owners and observable acceptance criteria.
- Assess process riskConsider product impact, information integrity and recoverability.
- Review design fitMap proposed functions and settings; resolve material gaps.
- Approve the baselineNamed reviewers approve revisions and disposition outstanding issues.
- Verify intended outcomesExecute checks, retain evidence and investigate unexpected results.
- Control subsequent changeAssess impact, revise relationships and repeat affected checks.
03What to include in a usable manufacturing URS
A practical manufacturing URS is organised around the work the system supports and the conditions under which it will be accepted. The structure below is specification practice, adaptable to equipment, software or integrated systems—not a universal regulatory template.
- Intended use and boundaries: describe the process, products, users and sites served. State what is included, what is excluded and where responsibility passes to another system or manual activity.
- Roles and operating environment: identify operator, supervisor, quality, maintenance and administrator needs. Describe relevant utilities, space, cleaning conditions, network access and physical working conditions.
- Process functions and performance: describe required operations, process-approved operating ranges, throughput needs and workload assumptions. State units and the basis for each range. Derive response-time and availability expectations from operational needs rather than adopting generic targets.
- Records and interfaces: identify records created or consumed; direct detailed controls to data integrity and record behaviour. For each interface, name the source, destination, business event, payload meaning and responsible owner. For example, an MES–QMS integration needs an agreed meaning for the status exchanged. Leave detailed failure and recovery behaviour to the worked interface case.
- Availability, maintenance and dependencies: describe operating schedules, acceptable planned interruptions and access for servicing. Identify supporting manual procedures, facilities, instruments and upstream master data, including who provides and maintains each dependency.
Give every requirement a reviewable structure
For each requirement, record the need, rationale, accountable owner, classification, acceptance criterion and relevant dependency. Classify process constraints separately from preferences: a constraint defines an operating boundary; a preference expresses a desirable choice. Do not let a convenient implementation choice silently become a process constraint.
Make acceptance criteria describe observable results under stated conditions. Where a value remains unresolved, identify its decision owner and basis for resolution rather than inserting an unsupported limit.
04Writing a testable requirement: a weighing example
A practical requirement pattern is: actor or trigger; operating condition; expected behaviour; observable result; acceptance criterion. Describe what a reviewer can demonstrate, not merely an intention such as accurate weighing. Keep each requirement atomic: give it an independently assessable outcome. Related requirements can share a test script, provided the evidence distinguishes their results.
Define the process assumptions
Assume a manufacturing operation weighs net material mass in kilograms. The approved process definition supplies target mass T and non-negative tolerance t, both in kg. The captured mass m is also in kg. For this example, the process accepts stable readings at either boundary: the inclusive acceptance interval is T − t through T + t. Inclusive boundaries are an example assumption, not a regulatory rule. Populate T and t with site-approved process values; see weighing tolerance bands.
Example acceptance requirement: when the operator requests capture against an approved process definition, with a stable instrument reading explicitly identified as kg, classify the weighing as accepted if and only if T − t ≤ m ≤ T + t. The observable weighing result identifies the captured mass, unit, applicable process definition and acceptance status. Separate related requirements define missing-unit and unstable-reading handling.
Make the acceptance decision observable
For valid kg readings, exercise both boundaries and values outside them using the instrument’s supported resolution. Unless stated otherwise, assume the reading is stable.
| Input condition | Intended result | Observable evidence |
|---|---|---|
| m = T − t | Accept lower boundary | Weighing result records mass in kg and accepted status. |
| m = T + t | Accept upper boundary | Weighing result records mass in kg and accepted status. |
| m < T − t or m > T + t | Reject as outside tolerance | Weighing result records mass in kg and out-of-tolerance status. |
| Unit missing | Block acceptance; do not infer kg | Capture status identifies missing unit; no accepted result is created. |
| Instrument reports unstable | Block acceptance, even inside the interval | Capture status identifies instability; no accepted result is created. |
The user outcome is a defensible capture decision. Instrument drivers, database structure and screen layout belong in design specifications unless a genuine process dependency constrains them. For example, retaining compatibility with an installed instrument can be justified when its stability indication determines capture eligibility; document that dependency rather than prescribing a driver without a reason. Review the wording against common specification-review controls.
05Linking requirement revisions to design and acceptance evidence
In common configuration-management practice, a requirement keeps a stable identifier while its revision, approval state and classification are separate attributes. Avoid identifiers based on page numbers, section positions or categories that may change. This lets requirements traceability follow the same requirement through edits without confusing identity with status.
The data-model diagram below describes relationships, not a mandatory document format. Each link points to a specific revision or baseline rather than merely a document title.
These relationships can be many-to-many. A shared verification case may address several requirements; a single requirement may need configuration review, interface testing and operational evidence. A CSV traceability matrix is a view of those relationships, not proof that every linked item is suitable.
Keep a planned test reference distinct from its executed result. Execution evidence identifies the case revision, tested configuration, observations and outcome. A pass against an obsolete system baseline or requirement revision does not settle current acceptance: reviewers assess whether the evidence remains applicable and document reuse or further verification.
Use the diagram during review to expose missing links, superseded evidence and unresolved deviations. Preserve baseline relationships so that change control after approval can identify affected requirements and evidence rather than replacing the history with the latest populated matrix.
Requirement
Defines the controlled user need.
- requirement_id
- revision
- requirement_text
- acceptance_criteria
- owner_id
Design Mapping
Connects requirements to implementation decisions.
- mapping_id
- requirement_id
- requirement_revision
- design_ref
- design_revision
Verification Case
Defines conditions and expected results.
- case_id
- case_revision
- requirement_revision_refs
- preconditions
- expected_result
Evidence Record
Preserves execution context and outcome.
- evidence_id
- case_revision_ref
- system_version
- result
- executed_by
Change Record
Links approved changes to affected baselines.
- change_id
- affected_requirement_refs
- impact_assessment
- approval_ref
- baseline_revision
06US regulatory baseline: obligations, scope and specification practice
Use a URS as a practical bridge between applicable obligations and system acceptance—not as a substitute for either. Start with the product, jurisdiction, process and intended use. The mapping below concerns US finished pharmaceuticals under 21 CFR 211; keep food, dietary supplement, device, chemical and non-US applicability decisions separate.
| Jurisdiction and product scope | Binding provision | Requirement implication | Possible acceptance evidence |
|---|---|---|---|
| US finished pharmaceuticals; computer or related systems | 21 CFR 211.68(b): appropriate controls must ensure that changes in master production and control records or other records are instituted only by authorized personnel. | Proposed URS: define which roles may change each applicable record type. | Role-based tests showing permitted changes succeed and unauthorized attempts are blocked. |
| US finished pharmaceuticals; computer or related systems | 21 CFR 211.68(b): input and output of formulas, records or data must be checked for accuracy. | Proposed URS: identify inputs and outputs to check and their reference values. | Documented comparisons against independently established reference data, including units and calculation results. |
| US finished pharmaceuticals; computer or related systems | 21 CFR 211.68(b): the degree and frequency of input/output verification must be based on system complexity and reliability. | Proposed URS: identify the verification strategy and its complexity-and-reliability rationale. | Reviewed strategy and execution records showing the selected checks and frequency. |
The last two columns are implementation choices, not prescribed statutory methods. Role checks, reference-data comparisons and documented verification strategies need process-specific justification. Carry the selected controls into the functional requirement specification, then connect them to design and acceptance records through a CSV traceability matrix. A passing demonstration supports only the conditions actually tested.
For US finished-pharmaceutical equipment, check the current text of 21 CFR 211.63 for scope. For US electronic-record applications, check the current text of 21 CFR 11.10 for scope before mapping it to the system.
Keep evidence categories distinct: binding provisions establish obligations; official guidance offers recommendations; industry practice supplies specification and testing methods; software implementation evidence demonstrates particular behaviour. Guidance recommendations and vendor feature descriptions do not belong in the binding-provision column. Human review decides acceptance and operational release, as described under who signs below.
07Data integrity: specify record behaviour, not just principles
ALCOA+ is an industry-practice vocabulary for data integrity, not a standalone regulation. In a URS, translate its attributes into observable behaviour rather than asking a supplier to demonstrate an undefined state of compliance.
- Attribution and timing: identify the record owner, the user or device creating each entry, and how timestamps, time zones and delayed entries are represented.
- Readability and original information: retain original values, units and contextual metadata; specify how reviewers retrieve and interpret them.
- Accuracy and consistency: define transformations, calculation-rule versions, rounding and unit conversions, with checks that expose inconsistent representations.
- Completeness: identify the events and related records needed to reconstruct the activity, including reviewable changes showing previous values, changed values, actor, time and reason.
- Endurance and availability: define readable retrieval over the process-defined retention period, backup restoration and reconciliation against an identified recovery point.
Distinguish source data from calculated results, displayed values and exported representations. For example, a rounded display is not the original measurement. Specify how reviewers follow identifiers from a result to its inputs and transformation rules, and how exports preserve or reference that context. Consider whether dynamic or static records support the intended review.
Recovery success means more than files opening. Define acceptance for a usable, interpretable record set: relationships, metadata, change history and calculation context remain accessible; missing, duplicated or mismatched information is identified and reconciled before use. Apply a comparable testable-requirement pattern, documented in a functional requirement specification, to retrieval and restoration scenarios, naming the expected evidence and reviewer decision.
08Interfaces and exception recovery: a failed-transfer example
Assume a manufacturing system sends a material-consumption transaction to an inventory system: quantities are in kilograms, timestamps use UTC, and each business transaction has a stable identity. In this hypothetical MES–ERP integration, the receiver commits the inventory deduction, but its acknowledgement never reaches the sender. The sender knows it attempted delivery; it cannot infer whether the receiver committed, rejected or never received the transaction.
Specify the business invariant as an acceptance outcome: retrying the same consumption transaction does not create an additional inventory deduction. Leave the technical mechanism to the design specification; a particular middleware product or duplicate-detection technique is not the user need.
- Status enquiry: retrieve the receiver’s recorded outcome using the stable identity. If the outcome remains unknown, retain a visible pending state rather than treating a timeout as rejection.
- Retry and duplicate detection: preserve the business identity across attempts. Recognise an already committed transaction and return its outcome without another deduction; flag conflicting content under the same identity.
- Rejected payloads and stale master data: stop the affected transaction and dependent processing. Record the rejection reason; require correction and an authorised resubmission decision rather than blind repetition.
- Unavailable destination: retain the transaction as pending, with controlled retries and escalation. Specify which dependent activities remain blocked until inventory status is resolved.
- Partial processing: stop dependent processing and require a documented disposition. Establish which effects occurred before deciding whether to complete or reverse them.
- Reconciliation and manual correction: compare identities, materials, quantities and outcomes across systems. Link any authorised correction to the original transaction, preserving its history and rationale.
Acceptance evidence should include sender status, receiver inventory entries and message history with identities and UTC timestamps. Start a recovery test after receiver commitment with the acknowledgement suppressed—not merely from a clean failed send. Exercise enquiry and retry, then demonstrate reconciliation without an additional deduction. Assign unresolved outcomes and permission to resume to named reviewers and decision owners under the site's quality assurance process.
09Who reviews, approves and owns unresolved requirements
URS review responsibilities are a site-defined governance choice, not a universal statutory signatory list. A practical allocation separates authorship from specialist review, process ownership, quality review and technical assessment:
- Authors consolidate needs, resolve ambiguous wording and maintain the draft.
- Operators assess usability, task sequencing and foreseeable workarounds.
- Process owners confirm intended use, operating boundaries and business priorities.
- Quality personnel assess quality risks and alignment with the site's quality assurance process.
- Engineering assesses equipment suitability, maintainability and physical interfaces.
- IT assesses architecture, access administration, infrastructure and support dependencies.
- Validation contributors assess testability, acceptance criteria and the planned evidence.
At each computer system validation lifecycle gate, distinguish approval of the requirements baseline from acceptance of verification results and authorization for operational use. Record each decision separately, with the accountable role and decision basis.
Keep unresolved assumptions, dependencies and exceptions visible in an issue log or quality risk register. Assign an owner, intended disposition and closure evidence; identify any issue that blocks a decision. A signature should not silently turn an unresolved issue into an accepted outcome.
Software can support recording and routing decisions, but accountable people approve requirements, assess evidence and authorize release; software does not make a site compliant.
10Changing an approved URS: a calculation-rule example
Assume an approved calculation defines net mass as gross mass minus tare mass, with all quantities expressed in kilograms. The business request is to align stored and reported net-mass results under a revised rounding policy. In typical change control, that request starts an impact assessment; it is not yet an implementation specification.
Separate the requested outcome from its consequences: when rounding occurs, which precision and tie-breaking rule apply, what value is stored, what reports display, and what interfaces transmit. Define acceptance criteria for each affected behaviour, including agreement between stored, exported and reported values.
Identify affected URS revisions, calculation-design references, configuration settings and verification cases. Preserve the prior approved baseline together with its test results and acceptance decisions, so reviewers can distinguish previously accepted behaviour from the proposed change.
Make historical treatment explicit: whether existing results remain unchanged, whether regenerated reports retain the original calculation basis, and when new calculations adopt the revised rule. Any historical recalculation belongs in a separately assessed decision, not a silent update.
Use risk-based validation to select targeted regression checks around rounding boundaries, unchanged subtraction logic, reports and interface outputs. Review affected evidence rather than assuming identical testing for every change. Consult applicable obligations such as 21 CFR Part 211; accountable reviewers approve the revised acceptance basis.
11Common URS failures and the controls that expose them
In manufacturing specification reviews, prioritize ambiguity that could change process execution, record meaning or acceptance decisions. Separate necessary outcomes tied to applicable requirements, including obligations such as 21 CFR Part 211, from preferences that do not affect compliance or process execution. The following are practical review controls, not additional regulatory requirements.
| Failure | Operational consequence | Review control | Evidence of resolution |
|---|---|---|---|
| Conflicting departmental requirements | Production wants editing; quality wants finalized records protected. | Document a joint process decision: finalized records remain protected; corrections follow an approved route. Do not leave implementers to choose. | Approved decision linked to reconciled requirements. |
| Uncontrolled attachments | Accepted content changes without changing the URS. | Use document control to identify each accepted attachment and revision within the approved baseline, not merely a mutable file location. | Baseline manifest identifying retained attachment revisions. |
| Hidden manual work | System completion is mistaken for procedural completion. | Identify the system-to-operator handoff, responsible role and required completion evidence. | Walkthrough showing system output, operator action and separately recorded procedural completion. |
| Unowned dependencies | Delivery assumes an interface or utility nobody owns. | Assign an accountable owner and acceptance condition to each dependency. | Owner acceptance and linked readiness evidence. |
| Features without identifiable process purpose | Scope expands without a defensible acceptance objective. | Ask which process outcome each feature supports; remove or defer unjustified features. | Recorded disposition or a linked process need. |
Apply the earlier writing, traceability and recovery controls to the resulting revisions. Close each review finding with evidence, rather than treating discussion or an assigned action as resolution.
12Implementation checklist: from approved requirements to CSV evidence
Use these practical questions to identify evidence gaps and follow-up actions, not to calculate a compliance score. Apply them at separate decisions: approving the requirements baseline and authorizing operational use.
Before approving the requirements baseline
- Are intended use, process boundaries, exclusions and dependencies clear?
- Does each requirement have an owner and observable acceptance criteria justified by the process need or risk?
- Have applicable obligations been identified for the product, jurisdiction and system use?
- Are design gaps and unresolved decisions visible, with an owner, follow-up action and disposition before approval where necessary?
Before authorizing operational use
- Does the CSV traceability matrix connect accepted requirement revisions to implemented design and reviewed acceptance evidence?
- Does evidence address interface failures and recovery, not only successful transfers?
- Are unresolved deviations and operating restrictions explicitly dispositioned for the proposed use?
An approved baseline and an authorized operational-use decision remain the separate human approvals described above; neither substitutes for the other.
Keep accepted requirements useful after implementation. They provide reference points for routine operating checks, periodic assessment and supplier-change impact reviews alongside supplier qualification. At replacement, use them to identify retained process needs, migration expectations and requirements that no longer apply.
Frequently asked questions
Q.Can a controlled spreadsheet serve as the specification, or is a dedicated requirements tool preferable?+
A controlled spreadsheet can serve as a URS when reviewers can identify the approved version, track individual requirements, record decisions and prevent uncontrolled edits. Apply document control to the file and its approval history. A dedicated tool becomes preferable when concurrent editing, dependencies and requirements traceability become difficult to manage reliably. Choose based on the work involved, not document appearance. Neither format replaces named owners, review of ambiguous wording or human approval of the baseline.
Q.How can a supplier's standard requirements template be adapted without accepting its assumptions?+
Treat the template as a checklist of questions, not an accepted description of your process. For each proposed requirement, record whether it applies, whose need it represents and which assumptions need confirmation. Examine default roles, interfaces, operating conditions and exclusions with actual users. Move implementation choices into the functional specification or configuration specification where appropriate. Keep supplier limitations visible as gaps or constraints rather than rewriting the user need to match the available product.
Q.How should an operating legacy system be documented when its original requirements are unclear?+
Build a current intended-use description from observed workflows, records, interfaces and interviews with process owners. Separate confirmed needs from inherited behavior and unresolved assumptions; existing operation alone does not establish suitability. Use a data integrity risk assessment to prioritize gaps affecting records and decisions. Within CSV, assess existing evidence against the reconstructed requirements and plan additional evaluation where needed. Date and approve the reconstruction honestly rather than presenting it as the original specification.
Q.What belongs in the baseline when a prototype is still needed to resolve a user need?+
Baseline the understood user outcome, operating context, constraints and the question the prototype is intended to resolve. Identify unresolved choices explicitly, with an owner and a decision gate before dependent design commitments. Keep prototype observations separate from approved acceptance evidence unless their suitability has been assessed. Findings can inform design qualification, while change control records the resulting baseline revisions. Avoid an acceptance criterion that merely says the prototype will determine what satisfactory performance means.
Q.How should essential acceptance criteria be distinguished from desirable features?+
Classify a criterion as essential when failure would prevent the agreed intended use or leave an unacceptable risk without an approved alternative control. Record that reasoning in the quality risk register, including affected users and operations. Desirable features improve convenience or efficiency but do not determine acceptance for the defined use. Give both categories observable evaluation criteria, but agree how each affects acceptance before testing begins. Avoid letting supplier pricing or delivery pressure silently change the classification.
Q.Can a shared test or supplier test result support several user requirements?+
Yes, where the evidence demonstrates each mapped requirement under relevant conditions. Map individual test steps and results, not merely a whole protocol, and explain which acceptance criterion each result supports. For a supplier's factory acceptance test, review the tested version, configuration, environment, execution records and unresolved exceptions. Identify what remains to be evaluated at the site, especially interfaces and operating workflows. Shared evidence can reduce duplicate execution, but a broad mapping does not compensate for missing coverage.
Q.How can acceptance criteria be shared with a supplier without exposing confidential formulas or process details?+
Describe the observable behavior using synthetic recipes, substitute material identifiers or representative datasets that preserve relevant relationships and boundary conditions. Keep a controlled internal mapping to the confidential process, and assess whether substitution hides an important failure mode. As part of supplier qualification, agree access restrictions, permitted uses and handling of test records. Where realistic confidential inputs are necessary, consider supervised testing or restricted disclosure. Sanitized examples are useful only when they still challenge the behavior being accepted.
Q.Can the requirements baseline form part of a procurement contract without freezing every design decision?+
A practical approach is to attach an identified URS revision while distinguishing agreed outcomes and constraints from design choices left to the supplier. Describe the expected design verification evidence, review responsibilities, exclusions and acceptance process. Establish how proposed substitutions and requirement changes affect cost, delivery and approval. Keep exploratory options outside the committed baseline until agreed. Procurement and legal reviewers can then address contractual effect without turning every screen layout or configuration choice into a fixed user requirement.
Primary sources
Explore this topic
URS sits inside 2 overlapping topic clusters in our glossary. Every neighbour is one click away.
Electronic records, signatures, audit trail and ALCOA+ data-integrity principles.
URS-through-PQ lifecycle, GAMP 5 categorisation and CSA's modern alternative.
Want to see how URS could fit into your own records and workflows? Explore the related V5 pages or talk to our team about what applies to your operation.
