A drone delivery pilot is not a fleet. It is one route, a dozen or two orders and one specific question: can you get a small parcel to a customer faster or more cheaply than you do today, without upsetting anyone on the way? For a shop with one location and a few hundred online orders a month, that question is answerable in about two months and for a sum you can plan for in advance.

What follows is the version of the process that survives contact with reality — the paperwork, the route design, the checklist, and the arithmetic that tells you whether the second pilot is worth funding.

Start with a pilot you can actually finish

The most common failure is scope. A shop decides to "test drone delivery" and ends up with a vague project that nobody can declare finished. Narrow it down until the test has edges.

Pick one question and write it as a sentence: Can we deliver small spare parts to customers within two kilometres of the shop, within 45 minutes of the order, for less than our current courier rate? That sentence tells you what to measure and when to stop.

Reasonable guardrails for a first run: one launch point, one short corridor, 20 to 50 real orders, four to eight weeks, and a cancellation policy written before the first flight. A hardware shop testing small fittings, a pet shop testing bags of treats, a bookshop testing a Saturday drop — the product matters less than the fact that it is light, robust and not time-critical in a regulated way.

Set your failure conditions now, while you are still unemotional about it. "If fewer than half the planned flights complete, or if cost per delivered order stays above what we pay a courier, we stop and write up why."

Permits, pilots and insurance come first

Aviation rules vary by country and sometimes by region, and they change. The general shape is consistent: the aircraft must be registered, the person flying it needs the appropriate remote pilot licence, and operations beyond visual line of sight or over people generally need a specific authorisation. Some jurisdictions have lighter regimes for short, low-risk flights; others do not.

Three practical moves:

  • Talk to your national aviation authority early. Ask what category your intended operation falls into and what evidence an application needs. A conversation now saves a rewritten plan later.
  • Check local rules too. Noise, land use and privacy complaints are usually handled at municipal level, not by the aviation regulator.
  • Look hard at your existing business insurance. Many commercial policies exclude aviation activities entirely. Third-party liability cover for drone operations is normally a separate arrangement.

There is a shortcut worth considering: partner with an operator who already holds the licences, insurance and approvals, and buy the flights as a service for the pilot. You learn the same operational lessons without owning the compliance burden. If you go that route, get the insurance certificate and licence status in writing, not in a chat message.

For anything involving permits, liability or customer data, take advice from a qualified professional in your own jurisdiction. This is a map of the terrain, not a substitute for it.

Design the route, the payload and the weather rules

Map the corridor before you promise anything. What sits underneath it — schools, playgrounds, main roads, overhead lines, hospital helipads? Where does the aircraft launch from, and who controls that space? A rooftop with no lift access can kill a pilot on day one.

Then think about the receiving end, which is where most pilots quietly fail. A customer in a house with a driveway is straightforward. A customer on the fourth floor of an apartment block is not. Ask for a photo of the intended drop point before accepting an address into the pilot.

Match the parcel to the aircraft, not the other way round. Packaging for flight is often heavier and bulkier than what you would hand over at the counter, and every gram eats into range and time in the air.

Finally, write your weather rule down in advance: the wind speed, visibility and precipitation at which you cancel. Cold reduces battery performance, and so does a headwind on the return leg. Expect a meaningful share of scheduled flights to be cancelled, and plan your sample size so cancellations do not leave you with nothing to analyse.

The pre-flight checklist

  1. One-page pilot plan: scope, route, dates, success metric, budget ceiling.
  2. Licence and insurance documents received and filed.
  3. Written agreement with the operator covering liability, scheduling, data and who contacts the customer.
  4. Confirmed drop-zone details for every test address, ideally with a photo.
  5. Customer script: what is arriving, when, and what to do if something goes wrong.
  6. Go/no-go weather thresholds, agreed and dated.
  7. Fallback plan for cancelled flights — courier, store credit, or collection.
  8. A single log sheet: order time, dispatch time, arrival time, outcome, cost.
  9. An incident process with a phone number a customer can reach in under a minute.
  10. A review meeting booked before the pilot starts, so it actually happens.

Build your own cost model

Published figures are close to useless here because they bundle wildly different operations. Build the model yourself from three written quotes and your own time records.

Fixed costs of the pilot

Operator mobilisation or setup fee, software, staff training, extra packaging, insurance, any regulatory application fees, and the marketing needed to recruit test customers. Add the hours your own team spends coordinating flights — that is a real cost even when nobody invoices you for it.

Cost per delivery attempt

Flight fee or aircraft and pilot time, energy and battery wear, packaging, your coordination time, and the cost of failed or repeated attempts. Failed attempts are the line most models forget, and they are usually the reason the arithmetic looks worse at the end than on the spreadsheet.

The formula that matters

Cost per delivered order = (total fixed costs ÷ number of successful deliveries) + variable cost per delivery.

With 30 completed deliveries, fixed costs are spread thin and the number will look ugly. That is normal and it is not a reason to stop. Run the same formula at 100 and 300 deliveries as a thought experiment, then ask what volume your customer base could plausibly support.

When you request quotes, ask what is excluded: waiting time, weekend surcharges, reflights, standby days lost to weather, and who pays for damage to a third party's property.

Reading the unit economics honestly

Compare cost per delivered order against what you pay now — a courier drop, a staff member on a bike, or the cost of a customer collecting. Then compare it against the gross margin on a typical basket. If delivery costs more than the margin, you are buying something other than delivery: speed, novelty, or learning. That can be a legitimate purchase, as long as you name it.

Be careful with retention claims. Thirty orders cannot prove that drone delivery keeps customers. What a pilot of this size can prove is operational: whether flights complete, whether customers accept the handover, whether your staff can absorb the extra coordination, and whether the cost per order moves in the right direction as volume rises.

Before you commit: a four-week run-up

Week one: define the question, the metric and the budget ceiling. Week two: talk to the aviation authority and get three quotes from operators. Week three: choose the corridor, confirm ten drop zones, and draft the customer script. Week four: sign the agreement, rehearse a full order end to end, and run the checklist with a stopwatch.

Then fly small, log everything, and decide on evidence rather than enthusiasm. If the numbers work at a realistic volume for your shop, you will know. If they do not, you will have spent a modest amount to find out — and you will have a document that stops you repeating the experiment next year.

Where permits, insurance or consumer contracts are involved, confirm the details with a qualified adviser in your jurisdiction before the first flight.

Photo: MotazPhotography / Pixabay