A supplier may have made many windows without having controlled the particular system, size, glass, hardware and performance route proposed for the next order. Turn a broad promise into a bounded, testable capability statement.
Capability is not a personality trait and it is not a photograph of machines. It is the demonstrated ability of a named organization and production route to translate a defined configuration into controlled drawings, materials, processes, inspections and conforming output. The useful question is not ‘Can you make any system?’ It is ‘What evidence shows that this team, plant and supply chain can control this proposed family, including its difficult variants, under the required programme and timing?’ The answer should survive comparison, sampling and a small trial without pretending that pre-order due diligence is certification.
Method boundary: This guide adapts evidence-based audit and quality-planning ideas to buyer due diligence. It is not an ISO audit, factory certification, product approval, engineering validation, capacity guarantee or recommendation of any supplier. Comparable evidence remains subject to confidentiality, authenticity and configuration review.
1. Rewrite ‘any system’ as a configuration-specific claim
The phrase ‘we can make any system’ is commercially convenient because it avoids the variables that determine difficulty. An operable window is not defined only by frame material. Opening type, maximum leaf size, aspect ratio, glass mass, wind and water requirements, thermal break, mullion arrangement, hardware family, reinforcement, drainage, finish, tolerances, installation interface, test route and annual quantity can change the necessary design and production controls.
Ask the supplier to restate the claim in a capability matrix: products routinely produced; products previously produced but not current; products that require new engineering, tooling or external processes; and products not offered. Identify the legal factory operator and actual production site. If another company provides design, extrusion, coating, glass, hardware assembly or testing, name that dependency instead of hiding it inside ‘our factory’. Outsourcing can be legitimate, but it changes the evidence and control questions.
Write a proposed capability statement that could be disproved: ‘Plant X proposes to design-review and fabricate system Y, including opening types A and B, schedule size/weight range C, glass and hardware route D, and performance evidence route E, using listed internal and external processes.’ The narrower wording makes a useful review possible. It also lets the supplier say where evidence is limited without losing face.
| Variable | Project definition | Supplier's claimed route | Evidence needed |
|---|---|---|---|
| System and opening types | Series, fixed/operable/sliding/folding/other | Routine / adapted / new | Comparable drawings, BOM and production record |
| Envelope of size and mass | Min/max width, height, area, sash and glass mass | Controlled limits | Calculations, hardware limits, trials and inspection |
| Performance pathway | Named methods, ratings and acceptance authority | Existing evidence / new test / calculation | Applicable reports, listings and approved plan |
| Components | Profiles, glass, hardware, seals, reinforcement and finish | Approved sources and substitutes | BOM, specifications and traceability |
| Processes and sites | Internal and outsourced steps | Named plant/provider | Process map, competence and acceptance controls |
| Volume and programme | Release quantities and dates | Capacity allocation | Resource-loaded plan and current constraints |
2. Set criteria before asking for evidence
A folder review is weak when the criteria are invented after seeing the folder. Create the acceptance questions from the actual window schedule, specification, destination, contract strategy and risk register. ISO 19011:2026 describes evidence-based and risk-based approaches for management-system audits. The ISO 9001 Auditing Practices Group paper on evidence collection explains that evidence is evaluated against audit criteria and that absence of found evidence does not automatically prove nonconformity. For buyer due diligence, that translates into a fair rule: define what would support the claim, search proportionately, and distinguish ‘not demonstrated’ from ‘cannot perform’.
Criteria should cover technical interpretation, design control, configuration management, purchasing, fabrication, inspection, nonconformity, traceability, testing, packaging and project communication. Assign a consequence to each result. A missing historical acoustic report may mean a new project test is planned; a missing method for controlling a safety-critical hardware substitution may be a pre-production hold. Do not award points for document volume. One complete trace through drawing, BOM, purchase, inspection and final unit can be more informative than a hundred unrelated certificates.
ISO 10005 gives guidance for establishing, reviewing, applying and revising quality plans for a product, project or contract, whether or not the organization has an ISO 9001-conforming system. Use that idea to request a project-specific control plan, but do not label the buyer's checklist an ISO 10005 assessment or require certification to a guidance standard.
| Question | Minimum persuasive evidence | Possible result | Procurement consequence |
|---|---|---|---|
| Can requirements be interpreted? | Marked-up specification, query log and resolved design basis | Supported / limited / open | Technical clarification before quote freeze |
| Can the proposed configuration be engineered? | Controlled drawing, checks and competent approvals | Supported for sampled family only | No extension beyond reviewed envelope |
| Can materials and external processes be controlled? | Approved BOM/providers plus incoming/outgoing controls | Supported with named dependencies | Substitution hold and source disclosure |
| Can fabrication repeat the design? | Work instructions, fixtures, first-off and in-process records | Supported for sampled operation | Pilot or witness point before volume |
| Can output be verified? | Inspection/test plan, calibrated resources and acceptance records | Limited by method/sample | Agree new test or inspection gate |
| Can the volume and timing be met? | Current resource-loaded plan and constraint ownership | Separate capacity conclusion | Phased release, contingency or no award |
3. Use an evidence ladder, not a showroom impression
Rank evidence by how closely it connects to the proposed configuration and how independently it can be checked. Marketing statements and curated photographs provide context. A sample drawing or component catalogue shows possible knowledge. A complete comparable job record shows that processes were used. A controlled trial against the new project shows current execution on the chosen challenge. Witnessed or independently assessed evidence can add confidence, but its scope still controls the conclusion.
Authenticity, applicability and adequacy are separate. A real report for a different series is authentic but may not apply. A correct drawing may apply but be too incomplete for production. A polished sample may demonstrate workmanship while hiding unapproved components. Ask where the record originated, which revision was released, how it connects to material lots and inspections, and whether the supplier is permitted to disclose it. Redact prior-client information rather than encouraging confidentiality breaches.
Certificates sit inside the ladder, not above it. A management-system certificate can support organization-level controls within its scope, after holder/site/status verification. A product listing can support a product-programme claim within its rules. Neither replaces the configuration trace. The buyer should not imply that reviewing these inputs certifies the supplier.
| Level | Example | Useful inference | Major limitation |
|---|---|---|---|
| 1 — assertion | Sales email, brochure or verbal promise | What the supplier claims | No execution evidence |
| 2 — resource context | Team list, equipment, site view, software | Potential resources exist | Availability and effective control unknown |
| 3 — controlled method | Procedure, checklist, work instruction, approved provider list | A defined process exists | Use and effectiveness not yet shown |
| 4 — comparable record | Drawing-to-BOM-to-inspection trace from similar unit | Process previously produced a relevant output | Differences and current conditions remain |
| 5 — project challenge | Reviewed drawing, sample or first-off for awkward project case | Current team can execute the sampled route | Sample does not cover every family/volume |
| 6 — independent evidence | Applicable accredited test, product listing or witnessed inspection | Defined result within independent scope | No automatic transfer beyond that scope |
4. Trace one comparable unit from requirement to release
Ask the supplier to select the closest legitimate prior configuration and explain why it is comparable. Do not accept a famous project name as the evidence. Request a redacted trace: input schedule and specification; technical questions; approved drawing revision; calculation or system-owner review where required; BOM; purchase or batch identity; work instruction; first-off and in-process inspection; test or final inspection; nonconformities; packing; and release. The aim is to see whether the records describe one unit rather than several unrelated examples.
Then create a differences register. A prior unit may have the same frame but smaller sash, lighter glass, different hinge series, no restrictor, different drainage path, another plant or different performance target. For each difference, ask whether it is inside a controlled family, needs a calculation, needs supplier/system-owner approval, requires tooling or trial, or requires new testing. ‘Similar’ is the start of analysis, not the conclusion.
Look for unflattering records. A capability story with no questions, revisions, nonconformities or corrective actions may have been curated. The stronger evidence is not perfection but controlled response: the issue was identified, segregated, dispositioned by authority, corrected, verified and prevented from silently recurring. Do not ask for personal data or proprietary information beyond the decision need.
| Stage | Record sampled | Connection to verify | Difference from proposed unit |
|---|---|---|---|
| Requirements | Schedule/specification/query log | Opening mark and controlling revision | Size, performance, destination or interface |
| Design | Drawing, BOM and checks | Same unit ID and approved disposition | Series, operator, reinforcement or drainage |
| Supply | Approved source and batch/purchase record | BOM item to received material | Glass, hardware, seal or finish route |
| Fabrication | Work instruction and first-off record | Drawing revision at workstation | Equipment, fixture, tolerance or site |
| Verification | Inspection/test report | Method, unit identity and acceptance criterion | Sample, method or required value |
| Release | Nonconformity closure, packing and release | Only accepted unit moved forward | Packaging, traceability or logistics constraint |
| Difference | Potential effect | Authority needed | Evidence/action before release |
|---|---|---|---|
| Larger/heavier operable sash | Hardware load, deflection, operation and safety | Responsible designer/system/hardware authority | Checks, limits and physical trial as required |
| New glass build-up | Mass, edge support, thermal and safety route | Responsible designer/glass/system authority | Updated BOM, setting and applicable evidence |
| Different hardware | Load path, machining, locking and maintenance | Authorised technical reviewer | Compatibility evidence and first-off inspection |
| New external processor | Process capability, traceability and lead time | Supplier QA/project authority | Approval, incoming criteria and contingency |
| New performance target | Test or calculation applicability | Specifier/engineer/scheme authority | Approved evidence plan; no unsupported interpolation |
5. Choose an awkward trial that exposes the real interfaces
A supplier will naturally show an easy, attractive sample. The buyer should select a representative high-risk configuration from the actual schedule: a large operable leaf near the proposed limit, an asymmetrical coupling, a low threshold with drainage constraints, a heavy laminated glass build-up, a special hardware sequence, or a finish/interface combination that creates coordination risk. Do not select an impossible torture piece merely to make the supplier fail. The challenge must be commercially plausible and tied to the intended order.
Run the challenge in gates. First issue controlled inputs and request questions. Then review the drawing, BOM, limit checks, process map, inspection plan and evidence plan. Only after paper closure should the team make a sample, prototype or first-off unit appropriate to the risk. Identify who pays, who owns tooling and intellectual property, whether the sample can be destructive-tested, and how results affect order award. A visual sample cannot substitute for a specified performance test.
Observe the supplier's reasoning, not just response speed. Useful behaviour includes identifying contradictions, declining an unsupported rating, escalating to a system or hardware owner, marking assumptions and controlling revisions. A fast drawing that silently changes glass, hardware or dimensions is negative evidence even when the rendering looks professional.
| Gate | Supplier output | Buyer review | Pass does and does not mean |
|---|---|---|---|
| Input receipt | Acknowledged documents, revision and assumptions | Completeness and conflicts | Shows disciplined intake; not design approval |
| Technical questions | Prioritised query register | Risk recognition and authority route | Shows inquiry quality; not solved capability |
| Design package | Controlled drawing, BOM, limits and differences | Configuration and requirement trace | Supports sampled design route only |
| Control plan | Process, providers, inspections and holds | Risk-based adequacy and ownership | Plan is not proof of execution |
| Sample/first-off | Traceable physical unit and records | Workmanship, dimensions, function and exceptions | Not performance certification or volume proof |
| Test/review | Agreed test, inspection and corrective closure | Method, criteria and result | Bounded to configuration and conditions |
| Scale-up | Resource-loaded production plan | Capacity, bottlenecks and contingency | Forecast still needs monitoring |
6. Verify the process, people and external-provider chain
Equipment is only one resource. Ask who interprets specifications, who can approve drawings and deviations, who programs or sets fabrication, who inspects first-off units, who controls gauges, who manages glass and hardware changes, and who releases product. Sample competence through a task and record rather than relying only on résumés. Confirm deputies for critical roles. A capable engineer who is unavailable to the project is not effective assigned capability.
Map every externally provided process. Extrusion, thermal-break assembly, coating, anodizing, heat treatment, lamination, insulating-glass fabrication, special machining, testing and packaging may sit outside the proposed assembly plant. For each dependency, record the provider, specification, approval status, purchase controls, traceability, acceptance, nonconformity route, lead time and alternate source. Do not misrepresent outsourced capability as internal; do not assume outsourcing is inferior when the control is demonstrable.
Inspect change control. Ask for one recent component or process change and trace who identified affected products, reviewed technical and certification implications, updated drawings/BOM/instructions, controlled old stock, notified customers where necessary and verified the first new output. Broad capability claims often fail at the boundary between a design revision and production reality.
| Process | Named owner/deputy | Evidence sample | External dependency / hold |
|---|---|---|---|
| Requirement and code interpretation | [record] | Marked-up inputs and query disposition | Local professional or scheme authority |
| System/detail engineering | [record] | Controlled calculation/drawing review | System, hardware or glass technical support |
| BOM and purchasing | [record] | Approved source and change trace | Component availability/substitution |
| Fabrication/setup | [record] | First-off and work instruction | Tooling, maintenance and competence |
| External processing | [record] | Purchase specification and acceptance record | Named processor and contingency |
| Inspection/test | [record] | Method, calibrated resource and record | Laboratory or scheme limits |
| Nonconformity/release | [record] | Segregation, disposition and authorised release | Customer/design authority for deviation |
7. Test capacity and delivery separately from technical capability
A team can be technically capable and still lack available capacity. Conversely, an empty production line does not prove it can control a new system. Build a resource-loaded programme for the proposed order: engineering release, long-lead materials, tooling, samples, testing, extrusion or finish slots, assembly, glazing, inspection, packing and shipment. Ask what other committed work competes for the same people, machines or external providers. Commercially sensitive totals can be handled with ranges or third-party review.
Use observable constraints instead of headline monthly volume. The bottleneck may be technical drawing approval, an external coating slot, IGU curing, one corner-crimping station, a testing laboratory, final inspection or rack availability. Compare planned takt and yield with a recent representative period, but do not extrapolate a staged one-day demonstration into annual capacity. Record assumptions about shifts, overtime, recruitment, new equipment and subcontracting.
Create a ramp plan with stop conditions. A pilot batch can confirm drawing usability, material traceability, tooling, workmanship and inspection flow before the largest release. Define what happens when yield, response time, supplier delivery or test results miss the threshold. A trial should reduce risk, not become an informal production order without acceptance and payment rules.
| Constraint | Current evidence | Project demand/assumption | Control or contingency |
|---|---|---|---|
| Engineering release | Assigned team and recent throughput range | Drawing families and decision dates | Priority, deputy and approval freeze |
| Critical material | Confirmed source/lead-time evidence | BOM quantity and release schedule | Buffer, approved alternate or staged release |
| Bottleneck process | Observed cycle/yield and maintenance status | Required weekly output | Pilot, extra shift or honest limitation |
| External process | Provider allocation and acceptance history | Slots and batch sizes | Second approved route or schedule float |
| Inspection/testing | Competent people/resources and lab booking | Lot/test frequency | Witness plan and no-release hold |
| Packing/logistics | Rack, weight and loading evidence | Shipment sequence | Reusable rack cycle and damage contingency |
8. Write a capability verdict with an envelope and expiry
Avoid a single supplier score. State what is supported: named product family, opening types, size/mass envelope, component route, plant, processes sampled, evidence dates and remaining external dependencies. State what is limited or excluded. Separate technical design capability, manufacturing process control, product/performance evidence, available capacity and project communication. A supplier can receive different dispositions for each.
Set conditions and revalidation triggers. Conditions might include completing the awkward trial, closing three design queries, obtaining an applicable product listing, approving an external processor, witnessing first-off inspection or limiting the initial batch. Revalidate after major design, plant, component, process, ownership or key-person changes, long production gaps, serious nonconformity or certification status changes.
The honest verdict is temporary and sampled. It does not say the factory can make ‘anything’; it says what the evidence supports today and what must happen next. That is more useful to both buyer and supplier because it turns capability into a controlled project decision.
| Dimension | Disposition | Supported envelope | Condition / recheck trigger |
|---|---|---|---|
| Technical interpretation/design | Supported / limited / open | Named families and sampled difficult cases | Resolve listed queries; recheck after design change |
| Production process control | [record] | Named plant/processes and trial revision | First-off witness; recheck after site/process change |
| Product/performance evidence | [record] | Exact listed/tested configurations | New test or approved applicability decision |
| External-provider control | [record] | Named components/processors | Approval, traceability and substitution control |
| Available capacity | [record separately] | Dates, quantities and stated assumptions | Monitor bottleneck/yield and competing commitments |
| Communication/ownership | [record separately] | Named lead, deputies and authority | Recheck after key-person or escalation failure |
Safe verdict: Evidence reviewed supports a limited capability conclusion for the named factory, system family, configuration envelope and controls sampled. It does not certify unreviewed variants, future capacity or project compliance.
FAQ
Does a large factory prove window-system capability?
No. Size and equipment are resource context. Capability for the proposed system requires configuration-specific design, process, people, supply, inspection and output evidence.
Is an ISO 9001 certificate enough?
No. Verify its entity, site, scope and status, then treat it as organization-level management-system evidence. It is not product certification and does not prove the proposed configuration, performance or capacity.
Can a past project photo prove experience?
It can identify a lead for further review. Request a lawful redacted trace showing requirements, drawings, BOM, production and inspection for a genuinely comparable unit, then document differences.
Should the buyer demand an on-site audit?
Use a risk-based sequence. Remote document and video review can identify questions; a targeted site visit, witness point or independent assessment may be appropriate when material risks remain. None should be mislabeled as accredited certification.
What is the best sample to request?
Choose a commercially plausible high-risk configuration from the actual schedule. It should expose important interfaces without being an artificial impossible challenge. Define what the sample can and cannot prove.
Does passing a sample mean the supplier can deliver volume?
No. A traceable sample supports the sampled execution route. Volume needs a separate resource-loaded capacity and ramp review, including current competing commitments and bottlenecks.
Are outsourced processes a failure?
Not inherently. Name each provider and verify specification, approval, traceability, acceptance, change and contingency controls. Do not describe outsourced work as internal capability.
How should a buyer record ‘not enough evidence’?
Use ‘open’ or ‘not demonstrated within this review’, identify the exact missing evidence and release consequence, and avoid converting absence of evidence into a claim that the supplier cannot perform.
Continue the project review
Connect this guide with the proposed window and door systems, then review how Cavorlen approaches manufacturing and quality control. When your schedule is ready, send it through the project review form.
Official sources used
ISO 19011:2026 — Guidelines for auditing management systems ↗ISO 9001 Auditing Practices Group — Evidence collection ↗ISO 9001 Auditing Practices Group — library and disclaimer ↗ISO 10005:2018 — Guidelines for quality plans ↗ISO 10013:2021 — Guidance for documented information ↗ISO — Guidance on ISO 9001 documented information ↗ISO/CASCO — Attestations of conformity ↗IAF CertSearch — Scope of accredited management-system database ↗FGIA — Product certification programmes ↗FGIA — Gold Label Certification Programme ↗NFRC — Certified Products Directory ↗Related buyer guides
This guide does not verify or imply Cavorlen capability, certify a factory, approve a window system, assign performance values, validate proprietary designs or guarantee production capacity, quality or lead time. ISO certification, equipment lists, samples, site tours and prior projects are inputs with bounded evidential value; none alone proves capability for the proposed product or project. Engineering, scheme and regulatory decisions remain with competent authorised parties.