A free shipping threshold is a discount, and it is one of the few discounts a store gives away without ever pricing it. Every order that crosses the line collects it, including the orders that were always going to cross. The only orders that pay for it are the ones that got bigger because the line existed.
That is the whole trade. You give up the shipping cost on a large group of orders in order to grow a smaller group of orders. Whether it works depends entirely on where you draw the line, and most lines are drawn by copying whatever felt normal or by rounding the average order up to a tidy number.
We think about the line as a paid promotion with a running cost, because that is what it is. Once you can say what the promotion costs you each month, you can decide whether it is earning its place, and you can move it with something other than a hunch.
Price the giveaway before you argue about the number
Start with what shipping actually costs you to fulfil, not what you charge. Take last month's orders, split them into the ones above the line and the ones below it, and multiply the count above the line by your average fulfilment cost for those orders. That figure is what the threshold cost you last month.
A labelled example. Say you had 1,000 orders, 600 of them above a $75 line, and shipping those 600 costs you an average of $9 each. The threshold cost you $5,400 in the month. Now ask the second question: how many of those 600 orders were sitting below $75 before the customer added something to cross it? If the answer is 50 orders, the promotion cost $5,400 to move 50 orders up by an average of, say, $18. That is $900 of extra revenue bought for $5,400 of shipping.
Those numbers are invented for the arithmetic, but the arithmetic is yours to run and it takes an afternoon. Most stores have never run it, which is why the line usually sits in the wrong place and nobody can say by how much.
Every order that crosses the line collects the discount, including the ones that were always going to cross.
Where the line goes when you set it from data
Set the line a little above the order value most of your customers naturally reach, not above your average. Averages are dragged upward by a handful of large orders and will place your line somewhere nobody was ever going to get to.
Use the distribution instead. Export your orders, sort by value, and find the point where a large group of orders clusters. The useful line sits just above that cluster, close enough that adding one more item feels like a small step rather than a second purchase.
The distance matters more than the number. If your typical order is one item at $40 and your line is $100, the customer has to nearly triple the order, so they do not try. If the line is $55, one small add-on gets them there and you have bought a genuine increase. The line should ask for an addition, never for a second decision.
This is also why one line rarely fits a whole catalogue. A store selling both $12 refills and $200 machines is asking two different customers to jump two different distances with one rule.
The carts sitting just below the line
The most informative report in this whole discussion is the count of orders that finish just under the threshold. Those are people who saw the offer, did not take it, and paid for shipping or left.
If that group is large, your line is too far from where people naturally land, or the store is not telling them how close they are. Both are fixable, and the second one is free. A cart that says how much more is needed, in money rather than in a progress bar nobody reads, does most of the work. Saying it on the product page as well as in the cart does more, because the decision to add a second item is made while browsing, not at checkout.
If that group is small and the group well above the line is large, your line is too low. You are giving away shipping on orders that never needed the encouragement, and you can move the line up and watch what happens to the count below it.
Either way, the report to build is a simple histogram of order values with the line drawn on it. It answers most threshold arguments in one look.
What to offer instead when the maths does not work
Some catalogues cannot support a threshold. Heavy or bulky products, wide price ranges, or thin margins on the most popular items all break the trade, because the shipping you give away costs more than the behaviour it buys.
In those cases we look at other ways to reward the bigger order. Flat rate shipping that is honest and cheap removes the surprise without giving the cost away entirely. A shipping cost shown early on the product page removes the shock that abandons carts, which is most of what a threshold was quietly doing for you. Free shipping attached to a bundle or a multi-pack pays for itself, because the order composition is fixed and you can price the shipping into the bundle.
Subscriptions are the cleanest version of this. Free shipping on a recurring order is affordable in a way that free shipping on a single order is not, because the second and third deliveries carry no acquisition cost.
The principle stays the same. Give the shipping away where it changes a decision, and charge for it where it does not.
Changing the line without guessing
Move the line in one step, then leave it alone long enough to read. A month is usually enough for a store with steady order volume, longer if your weeks are lumpy.
Watch four numbers together, because any one of them alone will mislead you. Average order value should rise. Items per order should rise with it, because if the average went up on the same number of items you changed price, not behaviour. Order count should hold, since a line set too high pushes people away entirely. And gross margin per order, after shipping cost, should be higher than it was, which is the only number that decides whether the change was worth making.
If order count falls while average order value rises, you moved too far. Step back halfway rather than all the way, and read it again.
Announce the change plainly on the site while it settles. A threshold that quietly moves upward reads to a repeat customer as a price increase, and they are not wrong. Telling people is cheaper than losing them.