Ground handling has operated for decades on industry standards rather than aviation law. IATA manuals, ISAGO, whatever each airline wrote into its contract. It worked, mostly, and it varied enormously between one apron and the next.
That’s ending. The EU has published the first ground handling safety regulations — Delegated Regulation (EU) 2025/20, which sets requirements for ground handling organisations, and Implementing Regulation (EU) 2025/23, which covers how competent authorities oversee them. They apply from 2028.
Three years sounds like plenty. It isn’t, and the reason has nothing to do with safety culture.
What “self-declaration” actually shifts
The framework doesn’t certify you in advance. You submit a declaration to your competent authority stating what services you provide, at which aerodromes, and that you comply.
That sounds lighter than certification. In practice it’s heavier, and here’s why.
Under a certification model, you prepare for an audit. There’s a date. Somebody assembles the binder, chases the missing signatures, and gets through the week. Everyone in this industry has done it.
Under a declaration model, you’re making a continuous assertion. There is no binder week, because there is no date to prepare for.
The authority can turn up at any point and ask you to substantiate it. The evidence either exists on the day they ask or it doesn’t.
So the question stops being “can we pass an audit” and becomes “can we produce, on a random Tuesday, the records that show what we asserted was true.”
For most ground handlers I’ve talked to, the honest answer to that is probably, eventually, if you give us a week and Zoltán is in.
What you’ll be asked to produce
The regulation covers a management system, safety management, personnel competence, occurrence reporting, and how you manage the interfaces between yourself, the aerodrome operator and the aircraft operators you serve. Read the actual text — this isn’t legal advice and I’m not a compliance consultant.
But strip it to what has to physically exist somewhere, and a pattern shows up. Nearly all of it is a record about a person, a task, or an event.
What has to exist somewhere
- Who is appointed to which accountable role, and since when.
- Who holds which qualification, when they got it, when it expires, who signed off the on-the-job component.
- What recurrent training each person has had and when the next one is due.
- What happened when something went wrong, what the root cause was, and what changed as a result.
- How you identified a hazard, assessed it, and what you did about it.
None of that is exotic. Every one of these facts exists in your operation today. The problem is where.
The folder problem
Here’s the shape of it in a typical operation.
Qualifications live in a spreadsheet the training coordinator maintains. Actual shift assignments live in the rostering system. Whether the person assigned to a pushback at 06:40 held a current certification at 06:40 is a question you can answer — by opening both, matching names by hand, and hoping the spreadsheet was updated after last month’s recurrent session.
Incidents live in email, or in a form that generates a PDF that goes in a folder. The corrective action that followed lives in the memory of the duty manager who implemented it.
And the thing that would actually prove the work was done to standard — that the task happened, at that time, by that person, in that sequence — usually doesn’t exist at all. It happened. Nobody recorded it.
This is fine right now, because nobody asks. From 2028 someone can ask.
Why this becomes a scheduling problem
Here’s the part that most compliance write-ups miss, and it’s the one that matters operationally.
A qualification isn’t an HR record. It’s a scheduling constraint.
If your rostering system doesn’t know who’s qualified for what, it will cheerfully produce an assignment that a person isn’t licensed to perform. The supervisor catches it, usually, because the supervisor knows their people. Then the supervisor is off sick, or the assignment gets swapped at 05:50 to cover a no-show, and nobody catches it.
You’ve now got a gap between the plan and what you’re permitted to do — and no record either way.
Handle qualifications as data the planner reads, and two things happen at once. The roster stops generating those assignments, because they’re infeasible by construction. And the record of who was assigned to what, under which qualification, is a by-product of planning rather than a separate administrative task.
That second part is the whole argument. Compliance evidence you have to go and assemble is expensive and always slightly out of date. Compliance evidence that falls out of running the operation costs nothing and is always current.
What I’d do with the three years
Not “start a compliance project.” Start with the gap analysis and be ruthless about what it tells you.
1. Pick five things the regulation requires you to prove. Try to prove them today. Not in principle. Actually go and do it. Pick a specific person, a specific shift last month, a specific pushback, and produce the evidence that they were qualified and it was done properly. Time yourself. That number is your real starting position, and it’s usually a shock.
2. Find out which records exist as data and which exist as documents. A PDF in a folder is not a record you can query. It’s a record you can lose. Anything that needs to be produced on demand should live somewhere queryable, and the distinction between the two is the actual scope of your work.
3. Fix qualification data first. It pays twice. It’s compliance evidence, and it’s what makes the roster stop producing assignments people aren’t licensed to perform. The failure mode here is also the worst one on the list. Everything else can wait a quarter. This can’t.
4. Start capturing execution now, even crudely. Three years of records is worth having in 2028. Zero years of records is a much worse conversation. Whatever you can instrument cheaply — a scan, a timestamp, a confirmation — start now. The data compounds and the tooling can improve later.
5. Decide who owns it. Compliance projects with no named owner do the same thing content calendars with no named owner do. Nothing, quietly, until the deadline.
Where we fit, honestly
We build optimization and execution-tracking systems for ground handling. We’re not a compliance product and we won’t sell you one.
What’s relevant is the by-product. OptiCheck holds people, skills, availability and tasks as structured data, which is what makes a roster respect qualifications rather than assume them. OptiScan captures what actually happened in the field. Between them you get a record of who did what, when, under which qualification — because the system needed that data to plan and to measure, not because someone was building an audit trail.
That doesn’t make anyone compliant. Compliance is a management system, and management systems are made of people and decisions. But a large share of what you’ll be asked to produce in 2028 is data about people and tasks. You can build that as a compliance exercise, or you can get it free from running the operation properly.
The second one is cheaper and it’s also just better operations.
Optimum Intelligence builds optimization and execution-tracking systems for ground handling. If you’re staring at 2028 and wondering where your evidence is going to come from, we’re happy to talk.