What should you actually charge for delivery?

Most delivery fees are set by copying the bakery down the road, which means a whole town can end up using a number nobody ever calculated.

A delivery fee is not income. It is a reimbursement, and it only reimburses you if it covers the trip. Anything it does not cover comes out of the cake. So the first question is not "what will customers accept", it is "what does one delivery cost me", and that number is almost always larger than the fee being charged.

All the arithmetic below is in ratios. Substitute your own currency and distances.

What one delivery actually costs

Four things, and only one of them is fuel.

The vehicle, over the round trip. The distance that matters is not the distance to the customer. A 12 km (7.5 mile) drop is a 24 km (15 mile) round trip, and on a single dedicated run you pay for all of it. Use your all-in running cost per distance: fuel, servicing, tyres, insurance and the share of the vehicle's value you are using up. Fuel is usually well under half of it.

Your time, door to door. This gets left out, and it is usually the largest line. Count loading and checking the order, the drive out, finding the address, parking, carrying, the handover, waiting when they are not ready and the drive back. On a 12 km (7.5 mile) drop in ordinary traffic that is comfortably an hour.

Here is the crossover, so you can settle it with your own numbers rather than guessing. On a 24 km (15 mile) round trip taking 65 minutes door to door, your time costs the same as your vehicle when your hourly rate is about 22 times your per-kilometre running cost, or about 14 times your per-mile running cost. Below that the vehicle dominates. Above it, your time does, and for almost every small bakery it does by a wide margin. A delivery fee set by watching fuel prices is set from the smaller half of the problem.

The deliveries that fail. Nobody home, wrong address, gate locked. If a fraction of attempts fail and each failure needs one more trip, the attempts per completed delivery are 1 divided by (1 minus that fraction). At 1 failure in 20 that is 1.053, so every trip should carry about 5.3 percent more than its own cost. If a failure also writes off the goods, which for anything perishable left on a doorstep in summer it does, add the full production cost at the same rate. That second part is often larger than the extra trip.

Packaging that only exists because it is travelling. The non-slip mat, the outer carton, the cold pack, the crush protection. A collection order uses none of it. That cost belongs to the delivery, and if it is buried in your product cost then your collection orders subsidise your delivery orders.

Getting the four inputs

None of the four can be borrowed from another bakery, and all four are measurable inside a week.

Running cost per distance. Take twelve months of vehicle spend: fuel, insurance, registration, servicing, tyres, repairs and depreciation, which is roughly what the vehicle lost in value over that year. Divide by the distance covered in the same period. If it is also your personal car, apportion by the share of distance that was business, not the share of time.

Door-to-door time. Time one ordinary drop and one awkward one, from the moment you start loading to the moment you are back and unloaded. Timed, not reconstructed afterwards, because the waiting and the parking are exactly the parts memory deletes.

Failure rate. Count first-attempt failures across your last fifty deliveries. Note separately how many of those wrote off the goods rather than only costing a second trip, because those two consequences are priced differently.

Delivery-only packaging. Buy one complete travelling kit and add up the line prices. It belongs to the delivery, not to the cake; the wider question of what else belongs on a quote is worked through in the costs most bakers forget.

What the arithmetic looks like once you have them

Take an entirely illustrative case, purely to show the shape of the answer. Call one hour of your time 100 units. Suppose the round trip costs 30 units of vehicle, the drop takes an hour door to door, the travelling packaging is 10 units and one attempt in twenty fails. That is 140 units before failures, and 140 divided by 0.95 is about 147 with the failures carried.

Every figure in that example is invented, and yours will differ on all four inputs. What survives the substitution is the structure: which line is largest, settled by the crossover above rather than by the illustration, and how the failure loading stays modest right up until a failure also destroys the cake, at which point it stops being modest at all. Once your own numbers are in, the line you were not expecting to be biggest is the one to price against.

The formula, and the one lever that beats it

For a single dedicated trip:

Run it once for a near drop and once for a far one. The two answers are further apart than any flat fee can bridge, which is the whole problem with flat fees.

Then the lever. All of that assumes one delivery per trip, and that is the expensive way to do it. On a run with six drops the return leg is shared six ways, and the only cost genuinely attributable to the sixth drop is the leg from the fifth plus the time at the door. The honest unit is cost per run divided by drops per run, and it falls fast as the run fills up.

This is why fixed delivery days beat delivering on demand, and it is a better lever than any fee structure. If you take one thing from this post: concentrating deliveries into fewer, denser runs cuts the cost per delivery further than any pricing change will.

Turning the number into a policy

A flat fee. Simple, instantly understood and wrong for every individual delivery. The near customer overpays, the far one underpays, and because the far one underpays, a flat fee quietly attracts the customers it loses money on. It is defensible when your area is genuinely small, your runs are dense and you have set it against your worst realistic trip rather than your average one.

Distance bands. Closer to the truth and easy to explain. Two things to get right. Cap the last band, so there is a distance where the answer is "we do not deliver there". And decide what happens at the edge, because a band boundary is a cliff: one extra street can move the fee a whole step, and that is where the arguments happen. A hybrid fixes the cliff. Charge a flat fee anywhere inside a radius, then add a per-distance rate on top beyond it. The fee is then continuous at the boundary and nobody gets a jump for crossing a line on a map.

Free over a threshold. This is a discount, not a delivery policy, and it comes out of profit. There is a clean test. Giving away a fee of F only pays for itself if the customer adds enough extra order value to cover it, and at a margin of m that means F divided by m in extra sales. At a 40 percent margin you need 2.5 times the fee in additional order value to break even. So set the threshold above your current average order value. Set it below and you are handing free delivery to orders you were already getting.

Also decide, deliberately, whether the threshold applies to distance-based pricing. The usual answer is no: the drive costs the same regardless of what is in the car. If your checkout waives a distance fee on large orders, you are giving your longest, most expensive trips to your best customers.

Six questions to ask of whatever prices it for you

If you cannot say what your system does here, that is the answer.

1. Is the distance one way or round trip? Almost every tool measures you to the customer. Your car does the trip twice.
2. When more than one rule could apply, which wins? Some systems return the cheapest valid option. If yours does, adding a cheap flat rate alongside distance pricing silently undercuts the distance pricing on every long trip.
3. What happens beyond your maximum range? Refusing the order and charging the top band anyway are both defensible. Not knowing which yours does is not.
4. What happens when the rule cannot be evaluated? A weight-based rule with no weight, a distance rule with an address it cannot resolve. The dangerous answer is zero, because zero looks like a valid price, and in any system that picks the cheapest option a zero beats every correctly priced alternative on the list. It should fail closed, not free.
5. Is the fee counted in your profit margin? It should not be. Delivery is a pass-through, so folding it in makes a long unprofitable trip look like a good day.
6. Does the fee cover any of your time? Most distance pricing prices distance only. If you want the hour covered it has to be in the rate you set.

What to do this week

• Time one real delivery door to door, including loading and the drive back. Write down the actual minutes, not the ones you would have guessed.
• Work out your running cost per distance for the whole vehicle, not fuel alone.
• Count last month's deliveries and the number of separate trips they took, then divide. That is your real drops per run, and it is probably lower than you think.
• Compare your current fee against the worst trip you accept.
• Then decide whether the fix is a higher fee or a tighter delivery day. It is usually the delivery day.

Which structure you land on should be decided by the shape of your map, not by preference. If everyone you accept is inside a short radius and your runs are dense, a flat fee set against your worst realistic trip is honest and takes no explaining. If your drops are spread out, the same flat fee is a subsidy running from your near customers to your far ones, and bands or the hybrid above are the fix.

Whichever you choose, write down the trip you set the fee against and keep it with the fee. That reference trip is the thing that goes stale. Fees are usually set while a business is small and local, and then quietly kept while the map grows around them, which is why so many bakeries are running a number that was correct two years ago.

The signal to revisit it is not a complaint about the price. It is a delivery you completed and would not accept again at that fee. Note that one down on the day it happens, and reprice against it, because the trip that hurts is never the average trip.

If you would rather not redo this on paper each time, delivery pricing is one of the things a bakery system should own: ibakepro prices it from the real driving distance to the customer's address and keeps the fee out of your margin figure, so a long trip cannot flatter a bad day. The doorstep time is yours to put into the rate.