A Part 21J programme can run for months on a change that everyone treats as minor — and then stop cold when the authority disagrees.
The drawings were revised. The analysis was updated. Approvals were collected. The package moved toward submission. Then someone — often during a readiness review, sometimes during the authority’s own sampling — asks the question that should have been asked at initiation: is this still a minor change?
If the answer is no, the consequences are not a paperwork tidy-up. They are delayed notification, reworked justification, possible reclassification of related work, and a schedule that was built on the wrong assumption about how much scrutiny this change would attract.
This article is about that failure mode: why major/minor classification drifts in design offices, what it costs, and how to make the decision early enough — and durable enough — that it survives contact with the authority.
Why Classification Drifts
Nobody sets out to misclassify a change. Drift happens because the decision is treated as a formality rather than a control.
Typical patterns:
- Criteria live in the handbook, not at the form. The design organisation handbook describes how to classify changes. The change request form offers a dropdown: Major / Minor. The link between the two is whoever fills the form remembering the criteria under time pressure.
- “It feels minor.” Classification becomes a gut call based on drawing effort, schedule urgency, or how similar the last change looked — not against documented criteria for airworthiness impact, certification basis effect, or means-of-compliance change.
- Classification after implementation. The change is designed first; classification is recorded when someone opens the compliance paperwork. By then the office has sunk cost and is biased toward the answer that keeps the current path open.
- Silent scope growth. What started as a local drawing correction accumulates related revisions, load-case updates, and substantiation changes. Nobody reopens the original classification when the package no longer matches the decision that authorised it.
- Informal consensus. A corridor conversation decides “we’ll run this as minor.” The record shows a classification with no rationale, no criteria cited, and no freeze against later revision of the change set.
Each of these is individually reasonable on a busy programme. Together they produce the late-discovery trap covered in designing for compliance: the programme discovers its true classification when discovery is most expensive.
What Major and Minor Must Mean Operationally
Labels are not the point. Decision criteria are.
For a Part 21J office, a usable classification regime answers, at the moment a change is proposed:
- What is the airworthiness / certification impact of this change?
- Does it affect the certification basis, means of compliance, or previously accepted substantiation in a way that requires authority involvement?
- Who is authorised to make that call for this project and change class?
- What evidence and impact assessment must accompany the decision?
- If the change grows, what triggers a reclassification review?
If those questions are only answerable by reading a handbook chapter and holding a meeting, classification is not operational — it is aspirational. Operational classification means the criteria are visible when the change is created, the decision is bound to a specific change set, and the record shows why, not only what.
Minor is not “small engineering effort.” Major is not “politically sensitive.” Those are programme instincts. Regulatory classification is about the effect on the approved design context and the communication path with the authority — and it has to be defensible months later under sampling.
What Getting It Wrong Costs
Misclassification rarely appears as a single line item. It shows up as delay, rework, and trust erosion.
| Failure mode | What it looks like | Why it was preventable |
|---|---|---|
| Late major discovery | Package treated as minor until readiness or audit | Classification was not forced at initiation against criteria |
| Notification lag | Authority should have been involved earlier | Minor path skipped the communication the major path requires |
| Scope creep without reopen | Original “minor” decision covers a larger change set | Classification was not frozen to a defined change baseline |
| Undocumented rationale | Auditor asks why it was minor; office reconstructs from memory | Decision recorded the label, not the criteria applied |
| Wrong approver path | Sign-off chain matched the wrong class | Classification and authority matrix were not linked |
| Parallel truth | Engineering plan assumes minor; compliance later assumes major | No single system of record for the live classification |
The expensive part is not rewriting a form. It is that every downstream plan — review depth, evidence depth, authority interface, milestone dates — was built on the wrong class. Fixing the label late means fixing the plan late.
Authority trust compounds the cost. An office that repeatedly surfaces major changes late looks like an office that cannot see its own design impact. That perception outlasts any single finding.
Practical Controls That Hold Up
Durable offices treat classification the same way they should treat approvals and baselines: as a property of the work system, not a parallel spreadsheet task.
Classify at Initiation
The change should not proceed into detailed design without a recorded classification. “We’ll classify it when we write the compliance summary” is how programmes discover majors three weeks before submission.
Initiation does not require perfect engineering completeness. It requires enough impact understanding to apply the office’s criteria — and a rule that incomplete impact assessment means the change stays in a holding state, not that it defaults to minor.
Bind Criteria to the Decision
The record should cite which criteria were applied and what the answers were. A dropdown without rationale fails the first audit question: show me why this was minor.
When criteria are structured — impact questions with yes/no or enumerated outcomes — the office can review consistency across programmes. When they are free text only, every classification is a one-off essay that nobody reuses.
Force an Impact Preview
Before classification is confirmed, the initiator should see what the change is expected to touch: requirements, compliance documents, evidence, open approvals, and related changes already in flight.
Impact preview does not replace engineering judgement. It removes the common failure where classification is made against a mental model of “this drawing” while the real change set already includes three substantiation revisions and a means-of-compliance tweak.
Freeze Classification Against a Change Set
Classification should be bound to a defined set of artefacts and revisions. If the change set grows past that freeze, the system should require a reclassification review — not quietly inherit the old label.
This is the configuration-control cousin of milestone baselining: the decision is only meaningful relative to what it covered.
Link Class to Authority and Workflow
Who may classify, who may approve the classification, and which approval path the change then follows should come from the same authority model that governs sign-off. A minor path that anyone can open, and a major path that requires named roles, only works if the class is enforced before those paths diverge.
Keep Reclassification Visible
Reclassification is not a failure — silent reclassification is. When a change moves from minor to major (or the reverse), the record should show who decided, when, against which criteria, and what programme actions followed (notification, additional evidence, schedule impact).
A Migration Path for Offices Still Deciding After the Fact
If classification today happens in meetings after design is underway, do not try to enforce a perfect model in week one. Sequence the change so the office learns before it is blocked.
- Inventory the last twelve months of changes. Sample major and minor labels against handbook criteria. The mismatches are your findings-in-waiting.
- Publish a one-page operational criteria sheet derived from the handbook — short enough to use at initiation, explicit enough to audit.
- Require classification + rationale at change creation in warning mode: allow progress, but flag anything without criteria answers or with incomplete impact notes.
- Add impact preview for high-risk change types (structures, systems safety, certification basis touchpoints) before full enforcement everywhere.
- Enforce freeze and reclassification triggers once the false-positive rate on flags is acceptable.
- Tie class to approval path and authority checks last — after the office trusts the classification data, not before.
Offices that reverse this order — hard enforcement on day one with weak criteria modelling — generate workarounds. Workarounds become the new informal classification regime.
Signals Classification Is Still a Form Label
- Changes routinely move from “minor” to “major” during submission readiness
- Classification rationale cannot be produced without interviewing the originator
- Related revisions are added to a change without reopening the class
- Programme plans assume minor until compliance says otherwise
- Engineers treat classification as paperwork belonging to someone else
- Authority questions about class are answered with reconstruction, not a record
If two or more of these are true, the office is carrying classification risk on every active change — and the risk peaks exactly when schedule margin is gone.
Final Takeaway
Major and minor are not administrative tags. They are early commitments about impact, scrutiny, and how the design office will communicate with the authority.
When classification lives in a handbook and a gut feel, programmes discover the truth late. When classification is made at initiation against visible criteria, bound to a change set, linked to authority, and reopened when scope grows, submission becomes confirmation of a decision the office already owns.
The practical test is simple: for any active change, can you show — without archaeology — what class it is, why, against which criteria, covering which artefacts, decided by whom, and what would force a revisit? If yes, classification is under control. If not, the next “minor” surprise is already in the programme.