No one can prove labels were checked

Examples are fictional local conventions. Unless another location is shown, the scope is DC01 / H1. Keep that scope with shortened identifiers.

When to use this guide#

Use this guide when an inspection has photographs but no object list, findings lack owners, or a report says labels passed without explaining what was examined.

What to gather#

Gather the population or sample list, naming-policy revision, previous findings, permitted observation methods, evidence location, and people responsible for review and correction.

Source context#

Sunbird's asset-audit note describes equipment audit logs with people and timestamps, alongside cabinet-position corrections. It illustrates an audit workflow, not a mandatory frequency or sampling rule. Sunbird asset audit application note.

Suggested method#

This is an editorial workflow to adapt to your site's approved process.

1. Define scope before inspecting: site, area, object types, population reference, and whether this is a full review or a sample. Record the selection method and exclusions.

2. Use consistent observation states: present/readable, missing, unreadable, mismatched, or not inspected. Record the object and criterion checked; a photograph without an identifier is difficult to reconcile later.

3. For each defect, capture evidence, the expected condition, and its operational context. Separate the observation from the proposed remedy so another reviewer can assess the finding.

4. Assign an owner and track the decision, correction reference, and due point. Retain the policy revision used; if an approved exception applies, record who accepted it and its scope.

5. Verify each claimed correction against its criterion. Record who checked it, when, and which evidence supports closure. Report inspected counts, open findings, and exclusions alongside the closed total.

Define the claim before walking the floor#

Write what the inspection is intended to establish. Presence, readability, naming conformity, physical association, endpoint accuracy, and record correspondence are different criteria. A readable cable ID does not prove the printed destination is current. A matching asset record does not prove the rear label was visible. Select the relevant criteria and retain their governing policy or requirement references before recording outcomes.

Define the unit being counted. An asset can have several physical labels, and a cable can have two end faces with different condition results. Decide whether the population contains objects, physical faces, or specific object-and-criterion checks. Use that unit consistently in the selection list and summary. Otherwise a report can count one cable as two passes in one section and as one finding in another without explaining the difference.

If the population is complete and available, record its revision and scope. If it is incomplete, state the limitation and describe the actual observed or selected list. Do not manufacture a full-population claim from a partial inventory. A bounded inspection can still be useful when its site, hall, object types, exclusions, and limits are clear.

Make sample selection reproducible#

Record whether the review covers the full stated population or a sample. For a sample, retain the actual selection list and describe how it was chosen. A selection focused on recently changed racks answers a different question from an arbitrary convenience walkdown. This guide does not prescribe a sample size, statistical confidence, or audit frequency. The site owner should choose an approach appropriate to its purpose and record what conclusions it can support.

Keep inaccessible objects in the accounting. Record them as not inspected for the relevant criterion, with the reason and follow-up responsibility. Replacing every difficult object with an easy one without retaining that history can make the final sample misleading. If the selection changes, preserve the original list, the revised list, and why the change was accepted.

Decide how previously known findings will be handled. A closure review of earlier defects is not the same as a fresh population inspection. Link prior finding IDs and name which original criteria are being rechecked. If both activities occur during one visit, keep their scopes and counts distinguishable so repeated observations do not accidentally become new unique inspected objects.

Sunbird's application note illustrates audit records with people, timestamps, and equipment-location corrections. It does not establish a required frequency or sampling method. The scope definitions, outcome states, and closure rules here are Rackstamp's editorial workflow. Sunbird asset audit application note.

Record the observation before deciding the remedy#

Identify the object and the physical face being examined. Capture the observed text as read, including incomplete or ambiguous portions, then record the independent evidence used to establish the intended object if needed. Do not rewrite an unreadable legend as a correct ID in the observation field merely because the register supplies the expected value. The difference between expected and observed is the finding.

Record outcomes per criterion. A label can be present and readable while mismatched to the record. It can also be unreadable with its physical association independently established. Use separate fields or linked observations so one favorable result does not erase another defect. For a criterion not examined, record not inspected rather than carrying forward a pass from a different check.

Capture enough context for another reviewer to locate and assess the condition. A close photograph may show damaged text but not the object or position. A wide view may show the location but make the text illegible. Use the site's permitted evidence method and link the observations deliberately. Avoid collecting unrelated rack or customer detail that does not help identify or assess the finding.

Describe operational context factually. Record whether the issue was discovered during an approved change, prevents the stated reading task, affects one face or several, or recurs in a known batch. Do not attach a dramatic severity label without the site's criterion for that judgment. The useful record identifies the practical consequence and lets the responsible owner prioritize it consistently.

Turn each defect into a verifiable assignment#

Write the expected condition and the observed departure, then assign an owner for deciding and completing the remedy. Keep the proposed action separate from the finding. A missing label may need identity investigation before reprinting; a mismatched label may require a records correction instead of physical replacement. The action should follow the evidence rather than the inspector's first assumption.

Give the assignment a due point or review trigger under the site's process and identify the evidence needed for closure. For a readability correction, that includes the complete readable installed legend and its object association. For a record mismatch, it includes the approved field decision and post-correction comparison. A work order requesting labels demonstrates assignment, not completion of either check.

Where the owner requests an exception, retain the actual authorized decision, its scope, reason, and any expiry or review condition. Record it as an approved exception rather than a physical correction. If the decision is only proposed, keep it open. A finding should not disappear from outstanding work because someone entered "exception" without an authority reference.

Field case: one object has two different criterion outcomes#

This fictional case occurs in DC01 / H1 for audit AUD-G30-301. The selected unit is a physical cable-end face. At cable CBL-G30-301, the inspector reads the complete cable ID from the normal approved viewpoint. The printed TO destination, however, differs from the current connection evidence under review. The inspector records readability as observed satisfactory and destination correspondence as a mismatch under finding FND-G30-301.

The initial summary draft counts the face as passed because its readability box is complete. The reviewer compares the summary with the criterion-level record and corrects that claim. The face was inspected for two criteria, with one satisfactory observation and one finding. The audit does not call the whole face conforming while the destination question remains unresolved.

The connection owner reviews the approved completion record and relevant endpoint evidence, then authorizes the corrected destination legend. The original observed text remains in the finding. After the replacement is installed through the site's work process, the verifier checks the full cable ID and the corrected destination against the approved relationship. The closure evidence names both the physical face and the criterion originally failed.

The final audit summary reports the inspected face population, the criterion outcomes, and one corrected finding without double-counting the cable as two separate objects. It retains the first readable-but-mismatched condition and the later correction. A future reviewer can understand why a photograph of clear text did not originally justify an overall pass.

Reconcile counts without concealing exceptions#

Reconcile the declared population with selected, inspected, excluded, and not-inspected quantities according to the stated scope. Where the inspection is a sample, keep unselected items outside the inspected pass total. If a selected face cannot be accessed, keep that limitation visible rather than silently reducing the denominator. Use the same unit defined at the start of the audit.

Reconcile findings separately from objects. One object may have several findings, and one bounded finding may cover several explicitly listed faces. State how grouped findings are counted. A summary of ten findings is not necessarily a summary of ten affected assets. Include enough explanation that another reviewer can reproduce the numbers from the register without inferring the counting convention.

Keep corrected findings, approved exceptions, and open findings in distinct totals. An exception can close an administrative decision while the physical departure remains. Expired or triggered review conditions need the site's follow-up process and should not remain buried in a historic closed total. Retain the authority and scope with the finding so the next audit can determine whether the exception still applies.

For repeat inspections, distinguish current observations from earlier closure evidence. A label corrected last month can develop a new defect; retain the prior history and link a new finding or reopened state according to the site's process. Do not overwrite the original verification date to make the earlier correction appear to have happened during the latest visit.

Review closure as a separate observation#

Ask the verifier to check the original criterion, not simply the proposed remedy. If the action was "reprint," confirm that the correct label is installed on the intended object and meets the relevant readability or correspondence requirement. If the action was a system update, confirm the approved value in the relevant current record and its agreement with the established physical evidence. Name the actual verifier and date.

Test evidence retrieval before finalizing the audit. A reference should lead to the observation, decision, and completed check needed to support the stated outcome. Broken links, unindexed image folders, or files available only to the original inspector can leave closure unreviewable. Fix the index or access arrangement through the responsible owner without claiming that an unavailable file was reviewed.

Use recurring findings to choose a specific follow-up investigation. Repeated corner lifting in one material batch calls for a material review; repeated destination mismatches after changes call for change-closeout review; repeated missing evidence calls for an inspection-record improvement. Link the process action to the relevant findings and track it separately. Correcting individual labels can be complete while the broader recurrence investigation remains open.

Worked example (fictional)#

Unless another site is named, the example scope is DC01 / H1.

SCOPE: 12 selected labels from 120; selection recorded
INSPECTED: 12   READABLE/MATCHED: 9   FINDINGS: 3

FINDING  CONDITION   CORRECTION   VERIFICATION
A-301    Unreadable  Reprinted    Closed; evidence EX-301
A-302    Mismatched  Proposed     Open
A-303    Missing     Assigned     Open

This fictional audit describes only its selected labels. It cannot support a claim that all 120 labels were checked.

Common mistakes#

• Calling an uninspected object a pass.
• Losing the naming revision used for the judgment.
• Closing a finding when correction was merely requested.

Verification checklist#

• The population, sample, and exclusions are explicit.
• Observations identify objects and criteria.
• Evidence references can be retrieved.
• Every closure names a verifier and date.
• Summary counts reconcile with findings and inspected objects.

Worksheet field dictionary#

Label audit and exception register. The example-value column shows one fictional worksheet row vertically.

Frequently asked questions#

How often should labels be audited?#

Set frequency through site governance, considering changes and earlier defects; this guide specifies no fixed interval.

Can an exception count as corrected?#

Record it as an approved exception with its authority and scope, preserving the distinction from a physically corrected label.

Can one object have both a satisfactory check and a finding?#

Yes. Record outcomes by criterion. A readable ID can coexist with stale destination text. Summarize the object and finding counts using the declared counting unit without turning a partial check into an overall pass.

What if a selected label cannot be reached?#

Record the relevant criterion as not inspected, retain the reason, and assign any required follow-up. If the selection changes, preserve that decision rather than silently substituting an easier label.

Can the original installer verify the correction?#

Use the independence and authority requirements established by the site's audit process. Whatever role performs the check, record the person, date, criterion, and evidence actually reviewed. Do not invent an independent review that did not occur.

More worked label examples

Two additional ways to apply this guide. Keep the exact identifiers and relationships consistent with your own approved records.

SCOPE: DC01 / H1; audit AUD-G30-101; policy NAM-G30-101 r3
POPULATION: 24 cable labels in POP-G30-101
SELECTED: 3 labels, selection list SEL-G30-101

OBSERVED FACE               CONDITION             FINDING
[CBL-G30-101]               Readable/matched      None
[CBL-G30-102]               Readable/matched      None
[CBL-G30-1??]               Unreadable ID         FND-G30-101

Third object's expected ID: CBL-G30-103, evidence EV-G30-103
Counts: population 24; inspected 3; not inspected 21
Inspected outcomes: 2 matched + 1 finding = 3
Open findings: 1; closed findings: 0
Claim supported: three selected labels were inspected

SCOPE: DC01 / H1; audit AUD-G30-201
FND-G30-201 / CAB-G30-201
Before face: [CAB-G30-2??]
After face:  [CAB-G30-201 | LOC DC01-H1-R201]
Correction CH-G30-201 -> final check EV-G30-201
State: CORRECTED; verifier/date recorded in audit register

FND-G30-202 / CBL-G30-201
Observed face: [CBL-G30-201], readable only from side access
Criterion exception: EXC-G30-201, authority/scope/expiry recorded
State: APPROVED EXCEPTION; physical condition retained

SUMMARY: 2 findings = 1 corrected + 1 approved exception
Corrected count: 1; exception count: 1; open count: 0
Exception remains subject to its recorded review conditions

Sources and applicability

• Sunbird: Asset audit application note ↗

  Manufacturer audit-log example; frequency, sampling, and closure workflow here are editorial.

  Read scope and linked guides

Planning guides and companion documents

• Choose a label printer for the work you actually do
• Choose barcodes, RFID, or AIM for the evidence you need
• Plan a legacy-room relabeling program
• Build a labeling budget and business case
• Data center labeling standards: build a project requirements map
• Labeling scope and handoff specification
• Label printer selection and trial plan
• Barcode, RFID and AIM technology pilot plan
• Legacy-room baseline, wave, and uncertainty planning template
• Measured labeling cost and benefit planning template

Related guides