How much should you charge for shipping?

A parcel that costs £3.49 to post rarely costs £3.49 to send. Once packaging, the minutes it takes to pack, the share that goes missing and the commission the platform charges on the postage you collected are all counted, the same parcel in the worked example below costs £5.50, and charging the £3.95 that felt about right loses £1.96 on every order — £2,352 a year at a hundred orders a month. This page works out what a parcel actually costs you, what you have to charge to recover it, why free delivery is a display decision rather than a pricing one, and the arithmetic that says whether a free-delivery threshold pays for itself or quietly funds itself out of your margin.

What one parcel actually costs you to send

Almost every seller prices delivery off one number: the postage label. It is the only cost with a receipt attached, so it is the only one that feels real. It is usually a little over half of what sending the parcel actually costs.

There are five blocks, and the last two are the ones that go missing. Fenwold Ceramics below is an invented business used to keep the arithmetic concrete; every figure in it is illustrative and belongs to the example only. Replace all of them with your own, which you can get from your carrier account, your packaging invoices and a stopwatch.

The handling line is the one people argue about, so work it rather than debating it. Four minutes a parcel at a hundred parcels a month is six hours and forty minutes of somebody’s month. If the delivery charge does not carry it, those hours were worked for nothing and the accounts have no way of showing it. Put in a rate you would actually accept from somebody else.

The failure line should also be worked rather than guessed:

Failure allowance per parcel = share of parcels that go wrong × cost of putting one right

Putting one right here means making and sending a replacement: goods £5.72 + packaging £0.86 + postage £3.49 + handling £0.93 = £11.00. Two in a hundred gives 0.02 × £11.00 = £0.22. If your own rate is one in two hundred the allowance halves. Use your figure, from your own records.

The sixth cost, which only exists on a marketplace

Most marketplaces charge their commission on the whole order, postage included. If you collect £3.95 for delivery on a platform charging an illustrative 10.5% of the order, £0.41 of that £3.95 never reaches you. You received £3.54 against a cost of £5.50.

So the shortfall is not the £1.55 it looks like. It is £5.50 − £3.54 = £1.96 per order. At a hundred orders a month that is £196, and over a year £2,352, on a business whose delivery line looked like it was roughly covering itself.

What you have to charge to actually recover it

Because the platform takes a percentage of the delivery charge as well as the item, you cannot simply charge what the parcel costs. Charging your cost guarantees you recover less than your cost. The charge has to be grossed up:

Delivery charge = cost to fulfil ÷ (1 − platform percentage on the order)

For Fenwold at 10.5%: £5.50 ÷ (1 − 0.105) = £5.50 ÷ 0.895 = £6.15.

Check it the long way, which is always worth doing once. The platform takes 10.5% × £6.15 = £0.65, leaving £5.50, which is exactly the cost. On your own site with a 2% card rate the same sum gives £5.50 ÷ 0.98 = £5.61, and with no percentage charge at all it is simply £5.50. The same parcel, three different honest charges, entirely because of who else is taking a cut.

Two things follow from that formula and both are worth saying plainly.

The gross-up grows with the percentage. At 10.5% you add 12p per pound of cost. At 25%, which is where a marketplace with advertising attached can end up, you add 33p per pound: £5.50 becomes £7.33. The delivery charge is not really a delivery charge on a high-commission channel; it is a delivery charge plus a tax on the delivery charge.

Rounding matters more than it looks. £6.15 rounded down to £5.95 costs you 20p an order, which is £240 a year at a hundred orders a month. Round up, not down.

Where the charge is capped by what buyers will accept

Sometimes the honest number is above what the channel will bear. That is a real constraint and the arithmetic does not override it, but it does tell you exactly what you are choosing. If you charge £3.95 when the honest figure is £6.15, you have decided to subsidise delivery by £1.96 an order, and the only correct response is to put that £1.96 into the item price instead of pretending it is not there.

On the item, at the same 10.5%, recovering £1.96 needs £1.96 ÷ 0.895 = £2.19 added to the price. An £18.00 mug becomes £20.19 with £3.95 delivery, rather than £18.00 with £6.15 delivery. The buyer pays £24.14 or £24.15 — the same total, within a penny of rounding. Which of those two the buyer prefers is a question about your customers, not about arithmetic — but the total has to be the same, and if it is not, the difference is coming out of your margin.

Free delivery moves the same money, as long as you move it

Free delivery is treated as a pricing strategy. It is not. It is a presentation choice about where the same money sits, and the arithmetic proves it exactly.

Hold the gross profit still at £6.00 a parcel and work the price both ways. Fenwold’s mug costs £5.72 to make and £5.50 to fulfil, and the platform takes 10.5% of whatever the buyer pays, postage included.

Identical, to the penny. The item price simply absorbed the £6.15 — £13.09 became £19.24, a rise of exactly £6.15. That is the whole mechanism, and it is why free delivery is neither clever nor dangerous in itself.

What it costs when you forget the second half

The damage comes from switching to free delivery without moving the price. Leave the mug at £13.09 and give the delivery away:

£13.09 − (10.5% × £13.09) − £5.72 − £5.50 = £13.09 − £1.37 − £11.22 = £0.50

Gross profit falls from £6.00 to £0.50, a cut of 91.7%. To make the same total gross profit you would need £6.00 ÷ £0.50 = twelve times the orders. There is no promotion in the world that multiplies orders by twelve, so the correct reading is that the offer cannot pay for itself and the only question is how long the business can fund it.

The one case where free delivery genuinely costs more

Some channels charge commission on the item but not on the postage. Where that is true, moving the postage into the item price makes it commissionable for the first time. Same arithmetic, one line changed: if the 10.5% applies to the item only, charging separately means the buyer pays £13.09 + £5.50 = £18.59, while free delivery at the same £6.00 gross profit needs a price of £19.24. The buyer pays 65p more, which is 10.5% of the £6.15 of postage that just became visible to the platform.

Sixty-five pence an order is £780 a year at a hundred orders a month. Before switching a channel to free delivery, check whether that channel charges commission on postage. If it does not, free delivery has a real price and it is 65p in this example.

The free-delivery threshold, and whether yours pays

“Free delivery over £40” is the most common version of the offer, and it is the one where the arithmetic is most often skipped. A threshold works when the gross profit on the extra spending it causes is bigger than the delivery you give away. That is one inequality and it has one useful rearrangement.

Minimum threshold = current average order value + (cost to fulfil ÷ gross margin on the extra spend)

Fenwold’s average order value is £22.00. Fulfilment costs £5.50. The gross margin on whatever the customer adds is 35%, taken from their own product margins after platform fees. So:

£22.00 + (£5.50 ÷ 0.35) = £22.00 + £15.71 = £37.71

A threshold of £40 clears that. A threshold of £30 does not, and the size of the miss is easy to see: an order lifted from £22 to £30 adds £8 of spending, which at 35% earns £2.80 of gross profit, against £5.50 of delivery given away. Every order that just crosses a £30 threshold loses £2.70. The offer works exactly as intended, customers respond exactly as hoped, and the business gets worse.

The bigger problem: the orders that were already over

The formula above prices one lifted order. It does not price the orders that would have cleared the threshold anyway, and those get free delivery for nothing at all. This is where most thresholds actually fail.

Say a month has 100 orders. Sixty were already over £40. The offer lifts twenty-five from £22 to £40, and fifteen stay below and pay for delivery as before.

The offer costs £310 a month, or £3,720 a year, and every individual piece of it behaves exactly as the theory says it should. The threshold is not too low. The problem is that 60% of orders were already clearing it, so most of the delivery being given away buys nothing.

Rearranged, the condition for the offer to break even is:

Orders lifted × lift × margin > (orders lifted + orders already over) × cost to fulfil

With a lift of £18 at 35%, each lifted order earns £6.30 and costs £5.50, netting 80p. Sixty orders already over cost £330. Breaking even needs £330 ÷ £0.80 = 413 lifted orders, which on a base of 100 orders a month is impossible. The honest conclusion is to raise the threshold well above the current average order value so that far fewer orders clear it by accident — at a £55 threshold, if only 20 orders were already over, the giveaway falls to £110 and the sum has a chance.

One flat rate across parcels that are not the same size

Most sellers do not send one parcel. They send several shapes, and a single flat delivery charge has to cover all of them. That is doable, and it is done with a weighted average rather than with the most common size.

Fenwold’s four bands, with the share of orders that falls in each, taken from their own despatch records:

Add the packaging £0.86, handling £0.93 and failure allowance £0.22 and the average parcel costs £5.70 to fulfil. Grossed up for the 10.5% platform charge, the flat rate is £5.70 ÷ 0.895 = £6.37.

Note that this is 22p above the single-parcel figure of £6.15 from earlier, purely because the mix contains heavier parcels than the one we costed. That is the flat rate doing its job: it over-recovers on the light ones and under-recovers on the heavy ones, and comes out level across the month.

The risk a flat rate carries, and how to size it

A flat rate is a bet on the mix, and the mix moves. Run a promotion on the heaviest product and the weighting changes underneath you. If the over-2 kg band goes from 5% to 20% and the lightest band falls from 45% to 30%:

(0.30 × £2.85) + (0.30 × £3.49) + (0.20 × £4.99) + (0.20 × £7.20) = £4.34

The average postage jumps by 65p a parcel while the flat rate stays where it was, so the business is now losing 65p on every order without a single number on any invoice having changed. At a hundred orders a month that is £780 a year, caused entirely by a successful promotion.

Two habits deal with it. Re-run the weighted average every quarter, or whenever the product mix changes materially — it takes five minutes with a despatch export. And if one band is far more expensive than the rest, take it out of the flat rate and charge it separately, because a single outlier band is what makes an average unstable. Fenwold’s over-2 kg band at £7.20 is more than twice the lightest; charged on its own, the flat rate for the remaining 95% falls to £6.16 and stops moving every time a heavy item sells well.

What this arithmetic cannot tell you

Everything above prices delivery correctly. Pricing it correctly and charging it successfully are different problems, and the second one is not arithmetic.

• It cannot tell you whether a higher charge loses the sale. This is the big one. The sums say £6.15 recovers your costs; they say nothing about how many buyers abandon at £6.15 who would have bought at £3.95. Only your own checkout data answers that, and the way to get it is to change one thing and watch what happens to total gross profit rather than to order count.
• It does not price returns. A returned parcel has been sent twice, handled three times and usually refunded in full, and the outbound postage is rarely recovered. If your return rate is material, that belongs in a cost of its own rather than inside the failure allowance, which here covers only losses and damage in transit.
• It does not cover sending abroad. International parcels carry different carriage, different handling time, customs paperwork and a materially different loss rate, and none of the figures above transfer. Cost those separately or do not offer them.
• Carrier prices change, and surcharges are not in the headline rate. Fuel, remote-area and peak surcharges can all sit outside the published price. Take your cost per parcel from what you were actually billed last month, not from a rate card.
• It cannot tell you whether the threshold caused the extra spending. The threshold sum above assumes the twenty-five lifted orders were genuinely lifted. Some of them would have spent more anyway, and counting those as caused by the offer flatters it. If you cannot separate the two, be pessimistic.
• The gross profit figure is not profit. Nothing here pays rent, software or your own non-packing hours. A delivery charge that exactly recovers its costs contributes nothing to those, which is an argument for charging the honest number rather than the comfortable one.

Once delivery is costed properly, it belongs in the unit cost and the price built on top of it — which is how to price a product so it actually makes money. And if you sell on a marketplace, check whether the commission applies to postage before using any of these figures: what Etsy takes from a sale and what eBay takes from a sale both work that line through.

Arithmetic and general information only — not financial, tax, legal or investment advice. Your own figures, and your own accountant, decide what any of this means for you.

Frequently asked questions

How much should I charge for shipping?

Work out the true cost first, then gross it up for whatever the platform takes. The cost is the postage label plus packaging plus the minutes it takes to pack, at a rate you would accept, plus a failure allowance for parcels lost or damaged. In the worked example on this page a 3.49 pound label is part of a 5.50 pound cost, and on a channel taking 10.5 percent of the whole order the charge has to be 5.50 divided by 0.895, which is 6.15 pounds.

Does free delivery cost me money?

Not in itself. If you move the delivery cost into the item price, the buyer pays exactly the same total and your gross profit is identical to the penny. The damage comes from switching to free delivery without raising the price. In the example here that turns 6.00 pounds of gross profit per order into 0.50 pounds, a 91.7 percent cut, and recovering it would need twelve times the orders.

What should my free delivery threshold be?

At minimum your current average order value plus the cost to fulfil divided by the gross margin on the extra spending. At a 22 pound average order value, a 5.50 pound fulfilment cost and a 35 percent margin, that is 22 plus 15.71, so about 38 pounds. But also check how many orders already clear the threshold, because those get free delivery for nothing, and if most of your orders are already above it the offer can lose money at any level.

Do marketplaces charge commission on the postage I collect?

Most do, and it changes what you have to charge. If a platform takes 10.5 percent of the whole order, the 3.95 pounds you collected for delivery is really 3.54 pounds received. Check your own platform: some charge on the item only, and on those the arithmetic for free delivery is different, because moving postage into the price makes it commissionable for the first time.

How do I set one flat delivery rate for parcels of different sizes?

Use a weighted average, not the most common size. Multiply each weight band by the share of orders that falls in it and add them up. In the worked example four bands give a weighted average postage of 3.69 pounds, which with packaging, handling and the failure allowance becomes 5.70 pounds of cost and a 6.37 pound flat rate after the platform gross-up. Re-run it quarterly, because a promotion on a heavy product moves the mix underneath you.

Other guides

• How much does Etsy take from a sale?
• How to build a cash flow forecast
• Agency utilisation rate: the formula, and the denominator that decides it