结论：审核看的不是你有没有报告，是你能不能证明报告覆盖了你在卖的产品

体系审核中，与检测相关的部分通常占相当比重。

很多企业的准备停留在「把报告找出来」的层面，但审核员关注的不止是报告本身。

审核员实际会追问的是：

报告覆盖了你在卖的产品吗？ 型号、规格、配置是否都被覆盖。

报告还有效吗？ 有没有过期、有没有因变更而失效。

变更之后重新验证了吗？ 设计、工艺、供应商变更后的处理。

日常的检验在做吗？ 进料、过程、出厂检验的记录。

检验的结论落实了吗？ 不合格品是怎么处理的。

能追溯吗？ 从产品能追到相关的检验和检测记录。

这六个问题串起来，就是审核在检测方面的主线。

审核员会看的文件

类别 具体内容
标准清单 适用标准及版本的识别
检测报告 型式检验报告及其覆盖范围
检验规程 进料、过程、出厂检验的作业文件
检验记录 各环节的实际记录
变更记录 变更及其影响评估
不合格记录 不合格品的处置
设备记录 检验设备的校准
供应商 供应商评价与来料证明

「检验设备的校准」是企业容易疏忽的 ——企业自己用的卡尺、量具、测力计也需要校准，而不只是实验室的设备。

常见的不符合项

报告覆盖不全。 系列产品中部分型号没有覆盖，或者报告的规格与实际销售的规格不一致。

标准版本过期。 依据的标准已换版而未重新评估。

变更未评估。 有变更记录但没有影响评估，也没有重新检验。

检验记录不完整。 有规程但记录缺失或流于形式。

记录与产品对不上。 检验记录无法关联到具体批次。

设备未校准。 检验用的量具没有有效校准。

不合格处置无记录。 出现不合格但处置过程没有记录。

「记录流于形式」是审核员容易看出来的 ——所有记录都是合格、数据完全一致、字迹和笔一模一样，这类记录的可信度会被质疑。

报告覆盖范围的准备

这是最需要提前理清的部分：

列出在售的全部型号和规格。 包括不同座宽、不同配置。

列出已有的检测报告。 报告覆盖的型号和规格。

做覆盖矩阵。 型号对报告，看哪些没覆盖。

对未覆盖的做说明。 通过典型型号覆盖的，要有覆盖论证。

典型型号的选取依据。 为什么这个型号能代表那些型号。

「典型型号的覆盖论证」是审核的常见追问点 ——用一个型号的报告覆盖一个系列，需要说明为什么可以覆盖，比如结构相同、材料相同、尺寸在覆盖范围内。

变更管理的准备

变更台账。 所有变更的记录。

影响评估。 每项变更对检测项目的影响分析。

处理结果。 需要重新检验的是否做了。

未重新检验的论证。 不需要重新检验的理由。

供应商变更的处理。 这一项最容易被漏掉。

建议把变更管理和检测报告的有效性绑定管理 ——每次变更时同步评估报告是否仍然有效。

日常记录的积累

审核前突击整理效果有限，记录的可信度来自日常的积累：

实时记录。 检验时当场记录，不事后补。

真实记录。 记录实际测量值，不填写标准值。

完整记录。 包括不合格的情况和处置。

签字与日期。 记录人和日期完整。

更改的规范。 记录有误时按规范更改（划改并签字），不涂抹重写。

「记录实际测量值」是可信度的基础 ——如果记录上所有数据都恰好等于标称值，审核员会怀疑记录的真实性。

内审的作用

提前发现问题。 内审是外审前的自查。

按外审的标准做。 内审松，外审就会紧。

重点查覆盖和追溯。 这两项是外审的重点。

做追溯演练。 随机抽一台产品做完整追溯。

整改闭环。 内审发现的问题要整改并验证。

「追溯演练」是有效的内审方法 ——随机抽一台产品，追溯到各关键件的来料、检验记录、相关报告，卡在哪里就是薄弱环节。

常见的准备不足

信息不全。 材料、工艺、使用环境的信息缺失，导致方案设计缺乏依据。

样品不具代表性。 试制样品或挑选样品，与量产状态有差距。

条件未明确。 测试条件没说清楚，数据的适用范围就不明确。

只看初始值。 忽略了环境和时间作用后的性能变化。

项目不完整。 遗漏了必要的项目，提交时才发现。

这五项在实际委托中出现频率较高，建议在项目启动时逐项确认，因为它们造成的返工成本远高于事前确认所花的时间。

数据的组织与保存

记录的完整性。 条件、方法、设备、样品状态、实测值。

可追溯性。 数据与样品批次、生产批次的关联。

保存期限。 覆盖产品的生命周期。

检索的便利。 需要时能快速定位。

版本的管理。 数据更新时的版本控制。

建议建立统一的数据台账，记录每组数据的来源、条件、适用范围和有效期。数据积累起来之后，后续的产品开发和变更评估都能受益。

与整体工作的衔接

这部分工作不是孤立的，它与几项内容相关：

与设计的衔接。 数据是设计决策的依据，也是设计验证的证据。

与风险管理的衔接。 识别出的风险点应当有对应的验证。

与变更管理的衔接。 变更时判断是否需要重新验证。

与生产控制的衔接。 关键参数的控制限值来自验证数据。

与售后的衔接。 失效分析需要原始数据作为对照。

建议在项目规划时就把这些衔接关系理清楚，避免各环节各做各的、数据无法互相支撑。

我们的做法

企业准备体系审核时，我们可以协助核对检测报告的覆盖范围 ——现有报告覆盖了哪些型号规格、哪些没覆盖、能否通过典型型号论证。

对于标准换版的情况，我们会说明变化的条款和是否需要补充检验，便于企业评估现有报告的有效性。

如果需要补充检验，我们会按缺口安排项目，而不是重复已有的有效数据，这样成本和周期都可控。

如果你在准备体系审核，需要核对检测方面的准备情况，可以直接联系：132 4819 8029。检测能力见服务介绍，产品分类见手动轮椅检测与电动轮椅检测，标准信息见标准查询。

English version

Conclusion. Testing occupies a substantial share of system audits. Many manufacturers prepare only to the extent of locating their reports, while auditors look beyond the reports themselves. What an auditor actually pursues is whether the reports cover the products being sold, across models, specifications and configurations; whether the reports remain valid, considering expiry and invalidation through change; whether re-verification followed changes to design, process or supplier; whether routine inspection is actually performed, across incoming, in-process and final inspection records; whether inspection conclusions were acted on, particularly how non-conforming product was handled; and whether traceability works, running from product back to the relevant inspection and test records. Those six questions strung together form the audit's main thread on testing.

Documents the auditor examines. The standards list covering identification of applicable standards and editions. Test reports covering type testing and its scope. Inspection procedures covering the working documents for incoming, in-process and final inspection. Inspection records covering what was actually recorded at each stage. Change records covering changes and their impact assessments. Non-conformance records covering disposition. Equipment records covering calibration of inspection equipment. And supplier records covering evaluation and incoming material certificates. Calibration of inspection equipment is readily overlooked: the manufacturer's own callipers, gauges and force meters require calibration too, not only laboratory instruments.

Common non-conformances. Incomplete report coverage, where some models in a series are uncovered or the report's specification differs from what is actually sold. Superseded standard editions, where the standard relied on has been revised without re-assessment. Unassessed changes, where change records exist without impact assessment or re-testing. Incomplete inspection records, where procedures exist but records are missing or perfunctory. Records that do not match products, where inspection records cannot be linked to specific batches. Uncalibrated equipment, where inspection gauges lack valid calibration. And undocumented disposition, where non-conformances occurred without a recorded response. Perfunctory records are readily spotted: every record conforming, data identical throughout, the same handwriting in the same pen, all invite doubt about credibility.

Preparing report coverage. This needs sorting out furthest in advance. List every model and specification on sale, including different seat widths and configurations. List the test reports held and the models and specifications each covers. Build a coverage matrix of models against reports, showing what is uncovered. Explain anything uncovered, with a coverage argument where a representative model is relied on. And state the basis for selecting the representative model, explaining why it represents the others. Coverage arguments for representative models are a common line of questioning: relying on one model's report to cover a series requires explaining why it may, on grounds such as identical structure, identical material and dimensions within the covered range.

Preparing change management. Keep a change register of every change. Assess impact, analysing each change's effect on test items. Record the outcome, covering whether re-testing was done where needed. Justify where re-testing was not done. And handle supplier changes, which are missed most readily. Bind change management to report validity so that each change prompts an assessment of whether the reports still hold.

Accumulating records day to day. Cramming before an audit achieves little, since the credibility of records comes from daily accumulation. Record in real time, at the point of inspection rather than afterwards. Record truthfully, entering measured values rather than nominal ones. Record completely, including non-conformances and their disposition. Sign and date completely. And correct properly where a record is wrong, striking through and initialling rather than obliterating and rewriting. Recording measured values is the foundation of credibility: where every recorded figure equals the nominal value exactly, an auditor will doubt the record's authenticity.

The role of internal audit. It finds problems early, serving as self-examination before external audit. It should apply external standards, since a lenient internal audit invites a strict external one. It should concentrate on coverage and traceability, the external audit's priorities. It should rehearse traceability on a randomly selected unit. And it should close the loop, with findings rectified and verified. Traceability rehearsal is an effective internal audit method: take a unit at random and trace each key component to its incoming material, inspection records and associated reports, and wherever it sticks is the weak point.

How we handle it. Where manufacturers prepare for system audit we can help check report coverage, establishing which models and specifications existing reports cover, which they do not, and whether a representative-model argument holds. For standard revisions we explain which clauses changed and whether supplementary testing is needed, so that the validity of existing reports can be assessed. And where supplementary testing is needed we scope it to the gap rather than repeating valid existing data, keeping both cost and duration under control.

If you are preparing for a system audit and want to check your testing preparation, get in touch: +86 132 4819 8029.