体系审核查检测记录时看什么

结论：审核员查的是链条，不是单份文件

检测记录的审核很少停留在「有没有这份报告」。真正的检查方式是选一个点，沿着链条往两头追。

比如从一份成品检验记录出发，往前追：依据的标准是哪一版、方法有没有确认、设备校准在不在有效期、操作人员有没有授权；往后追：数据怎么判定、不合格怎么处理、结论进了哪份放行记录。

链条上任何一环断了，这份记录的说服力就没了。所以准备的重点不是把文件做漂亮，而是保证每一环都能接得上。

常见的检查路径

第二行是最容易出问题的一条路径。 从设备往后追，要能说出这台设备测过哪些批次的产品。多数企业的记录是「产品—设备」的单向关联，反向查不了。

追到底会问什么

举一条完整的追问链作为例子：

审核员拿起一份成品检验记录，问：这个项目依据什么方法？——答：某某标准某条款。再问：这一版是现行有效的吗？——需要拿出标准清单和更新记录。再问：这个方法在本实验室确认过吗？——需要方法确认记录。再问：测量用的是哪台设备？——记录上要有设备编号。再问：这台设备的校准证书？——要能当场调出。再问：校准量程覆盖这次测量的范围吗？——需要比对证书和实测值。再问：操作的人是谁，授权了吗？——需要人员授权记录。

这条链上任何一环答不出来，整份记录的可靠性就要打问号。

记录里最常见的薄弱处

一是原始记录缺失或不规范。 只有整理过的报告，没有原始数据；或者原始记录有涂改而无签署确认。

二是标准版本失控。 记录上写的标准号没有年份，或者写的版本已经作废。

三是设备信息不完整。 记录上没写用的哪台设备，无法追溯到具体仪器。

四是判定依据不明确。 数据有，但依据什么判定合格没有写。

五是人员资质与实际操作不匹配。 授权做 A 项目的人签了 B 项目的记录。

六是修改留痕不规范。 数据修改用涂抹或覆盖，看不到原值。

这六项的共同点是：它们都不影响产品本身的质量，但都会影响记录的证据效力。

电子记录的特殊要求

越来越多的记录是电子形式，这带来额外的审核关注点：权限管理是否分级、修改是否留痕且可追溯到人、数据是否可导出为可读形式、备份与恢复是否验证过。

电子记录的修改留痕要求比纸质更严，因为电子数据修改后不留痕迹的技术门槛更低。如果系统不具备审计追踪功能，那么用它作为原始记录载体就存在缺陷。

迎审前的自查清单

建议按这个顺序自查：

随机抽三到五份检验记录，沿着前面那条链自己追一遍，看哪一环接不上；核对标准清单与实际使用的版本是否一致；检查所有在用测量设备的校准状态，特别是临近到期的；抽查人员授权范围与实际签署记录是否匹配；检查最近的不合格处置是否形成闭环；确认外部检测报告都有供方评价和接收记录。

这套自查做一遍，通常能发现两到三处需要补的地方，而这些如果留到审核时被指出，性质就不同了。

审核中的常见追问方式

除了顺着链条追，审核员还会用几种方式交叉验证：

时间线核对。 检验记录的日期与生产记录、设备校准日期、人员授权日期是否能对上。出现逻辑矛盾（比如检验日期早于设备校准日期）会被深入追查。

数量核对。 检验记录的数量与批量、抽样规则是否一致。

异常值追踪。 挑数据里的异常点，问当时是怎么判断的、有没有复测、复测记录在哪。

空白与遗漏。 连续编号的记录中间缺号，会被问缺的那些在哪。

这几种方式都不依赖专业技术判断，纯靠逻辑一致性，所以准备时要特别注意记录之间的相互印证。

纠正措施的闭环

审核中发现的问题要形成闭环：原因分析、纠正措施、有效性验证。检测相关的不符合项，有效性验证通常需要新的数据支撑。

只写「已加强培训」而没有后续的验证数据，这个闭环是不完整的，下次审核大概率会被翻出来。

记录的可读性

一个实际但常被忽略的问题：记录要让不熟悉产品的人也能看懂。审核员不了解你的产品，缩写、内部代号、只有本部门懂的简写，都会造成解释成本。

建议在记录表单上保留必要的说明栏，关键缩写有对照表。审核时需要额外解释的每一处，都是一次被深入追问的机会。

迎审当天的配合

准备做得再好，现场配合方式也会影响结果。几个实用建议：需要的文件事先整理到可快速调取的位置；指定熟悉记录体系的人陪同；被问到不确定的问题时如实说明并承诺查证，不要猜测作答；现场发现的问题当场记录，不要指望事后回忆。

如实说明比勉强解释好。 猜测作答一旦被发现前后不一致，影响的是整体可信度。

我们的做法

我们出具的报告会包含方法依据、试验条件和设备信息，就是为了让委托方在体系审核时能接上链条。报告上缺这些信息，委托方要么额外来问，要么在审核时解释不清。

对于需要完整追溯的项目，我们也可以在报告之外提供试验的原始记录副本，用于委托方的质量记录归档。这类需求建议在委托阶段提出，报告出具后再补办手续会麻烦一些。

如果你正在准备体系审核，或者想先自查一遍检测记录的完整性，可以把现有的记录样例发过来看，或者直接联系：132 4819 8029。能力范围见服务介绍，送检要求见送检要求，其他问题见常见问题。

English version

Conclusion. Auditing of test records rarely stops at whether a report exists. The actual method is to pick one point and follow the chain in both directions. Starting from a final inspection record, an auditor looks backwards at which version of the standard applied, whether the method was validated, whether the equipment calibration was current and whether the operator was authorised; and forwards at how the data were judged, how nonconformity was handled, and which release decision the conclusion fed into. If any link in the chain is broken, the record loses its evidential weight. Preparation is therefore not about making documents look good but about making sure every link connects.

Common audit paths. From a final inspection record, backwards to method, equipment and personnel, forwards to release decision and product disposition. From a measuring instrument, backwards to calibration and intermediate checks, forwards to which products it measured. From a nonconformity report, backwards to detection and acceptance basis, forwards to disposition, re-inspection and corrective action. From a design change, backwards to change review and verification plan, forwards to verification data and document updates. From an external test report, backwards to supplier evaluation and the commissioning document, forwards to receipt confirmation and how it was used.

The equipment path causes the most difficulty. Tracing forwards from an instrument requires knowing which batches it measured, and most record systems only link product to equipment in one direction.

How the questioning develops. An auditor picks up a final inspection record and asks what method the item follows; the answer names a standard clause. Is that version current? A standards list and update record is needed. Was the method validated here? A validation record is needed. Which instrument was used? The record must carry an equipment identifier. Where is its calibration certificate? It must be retrievable immediately. Does the calibrated range cover the measured value? The certificate and the reading must be compared. Who performed the work and were they authorised? An authorisation record is needed. A gap anywhere along that chain puts the whole record in question.

The most common weaknesses. Missing or irregular raw records, where only a tidied report exists or corrections are unsigned. Uncontrolled standard versions, where a standard number carries no year or refers to a withdrawn edition. Incomplete equipment information, preventing traceability to a specific instrument. Unstated acceptance basis, where data exist but the basis for the verdict does not appear. Personnel authorisation not matching actual signatures. And irregular correction practice, where values are obliterated rather than struck through so the original cannot be read. None of these affects product quality directly; all of them affect the evidential value of the record.

Electronic records. These attract additional attention: whether access rights are tiered, whether changes are logged and attributable, whether data can be exported in readable form, and whether backup and restoration have been validated. Requirements for change logging are stricter than for paper, because altering electronic data without trace is technically easier. A system without audit trail capability is deficient as a carrier of raw records.

A self-check before audit. Select three to five inspection records at random and follow the chain yourself to find where it breaks. Reconcile the standards list against the versions actually used. Check the calibration status of every instrument in use, particularly those near expiry. Sample personnel authorisations against actual signatures. Verify that recent nonconformities were closed out. Confirm that external reports all have supplier evaluations and receipt records. Working through this usually surfaces two or three items needing attention, which is a different matter from having them raised during the audit itself.

How we handle it. Our reports carry method basis, test conditions and equipment information so that clients can complete the chain during audit. Where a report omits these, the client either has to ask afterwards or cannot explain the gap on the day. For work requiring full traceability we can also supply copies of raw test records for the client's quality files; this is simplest to arrange when the work is commissioned.

Send us a sample of your existing records and we will review completeness. Phone or WeChat: +86 132 4819 8029.