Start with a testable hypothesis
"Trying out drones" is too broad an objective. A useful pilot starts from a hypothesis: given an alarm in a given area, an aerial mission can provide relevant information to the operator within the existing security procedure.
The hypothesis must define the event, the area, the time window, the expected information and the decision that information will help to make. It must also state its limits: weather conditions, people present, airspace restrictions, privacy and the alternative procedure when flying is not possible.
Choose a demanding but controllable case
The first site does not have to be the most spectacular. It should represent the future service without accumulating every complexity at once. A defined perimeter, identifiable events, verifiable connectivity and the collaboration of the security team provide better conditions for learning.
Aeronautical feasibility must be analysed from the start. In Europe, operations fall into the open, specific or certified categories depending on their level of risk. AESA states that the specific category covers medium-risk operations that cannot be carried out in the open category. If the mission fully matches a standard scenario, a declaration can be submitted; in other cases an operational authorisation and a risk assessment are required (AESA).
This classification is not a final formality. It shapes the aircraft, the training, the controlled ground area, the operating distance, the mitigations and the procedures.
Design the pilot as if it were already a service
Even if the scope is small, the pilot should incorporate from day one the functions that will make scaling possible:
- defined owners and activation criteria;
- normal, contingency and emergency procedures;
- integration with the alarm source and the operator's workstation;
- maintenance and availability checks;
- logging of flights, video, alerts and decisions;
- access control and a data retention policy;
- support, incident management and post-event review.
This avoids the "demo success": the drone flies in front of an audience, but nobody knows how to operate it from Monday to Sunday, who handles a breakdown or how it fits into the ARC.
Measure a baseline and a result
Without a baseline, any improvement will be an impression. Before deploying, it is worth recording how events are handled today: how many occur, how long it takes to obtain information, how many require a site visit, which areas create uncertainty and what data is documented.
Then comparable metrics are compared. An initial dashboard may include:
- Availability: share of time the service is ready.
- Execution: missions started, completed, cancelled and why.
- Information: time to useful imagery and events with additional context.
- Operation: technical alerts, support interventions and recovery.
- Compliance: missions within procedure, data access and incidents.
- Adoption: operator feedback and clarity when making a decision.
Universal promises of cost reduction or response in seconds should not be published on the basis of a limited test. First you need your own evidence, an explanation of the method and a clear distinction between measured results and projections.
Define continuity criteria in advance
The pilot needs an exit gate. Before starting, the thresholds that will justify expanding, correcting or stopping the project must be agreed. For example: minimum availability, share of useful missions, absence of serious incidents, acceptance of the flow by the ARC and documented feasibility to replicate it.
It must also be established what changes when moving from one to ten sites. Fleet management, support, supervision, updates, ARC capacity and operations coordination take on another dimension. Scaling is not copying the installation; it is standardising what is common without ignoring local conditions.
Learn and turn learning into product
Every activation should leave a lesson: a route that needs adjusting, an alarm that needs more context, a camera angle of little use or a manual step that can be integrated. Those observations must become controlled versions of procedures, configuration and training.
The market is evolving quickly. The European Commission expects civil emergency services, mapping, imaging, inspection and surveillance to be part of the European drone ecosystem by 2030, within the applicable legal frameworks (European Commission). The opportunity will belong to those who turn innovation into a repeatable operation.
A scalable pilot ends with a well-founded decision and a better manual than the one it started with. If it only produces an attractive video, it has tested an aircraft. If it produces data, procedures and integration, it has begun to build a service.