How to set a free shipping threshold that doesn't quietly lose money

The standard advice is to take your average order value and add 25 to 30%. It is repeated on nearly every page that ranks for this question, and it is incomplete in a way that costs real money: it models the orders that get bigger and ignores the orders that were already big enough. Those orders now ship free, and you were previously charging for them. That is the largest single term in the calculation and almost nobody includes it.

Why this is worth recalculating now

Shipping costs are moving again, and the direction is set for next January. FedEx announced its 2027 general rate increase on 18 September 2026: 5.9% on average across Express, Ground and Freight, effective 4 January 2027. It is the fourth consecutive year at that headline average, and the headline is not the number most retailers should plan against. Reported analysis of the published rate tables shows increases on packages from 1 to 20 pounds running between about 6.1% and 6.5%, with far zones reaching roughly 6.7%, and only parcels of 21 pounds and above easing back into the high fives.

Surcharges are rising faster than base rates, which matters more than the GRI for most e-commerce parcels. Additional Handling rises by roughly 6.8% to 7.6%, the residential Ground Extended Delivery Area surcharge rises about 9.1%, and the Oversize charge for zones 5 and 6 goes from $320 to $345 per package. If your average parcel is residential, and many are in a rural delivery area, your effective increase is meaningfully above 5.9%.

Retailers are already responding through the threshold. Deloitte's 2026 Retail Industry Global Outlook, based on 330 global retail executives surveyed in October and November 2025, reported that 67% expected to raise their free shipping threshold. Meanwhile shipping remains the single largest source of checkout abandonment: Baymard's research consistently finds extra costs such as shipping, taxes and fees are cited by around 48% of abandoners, against an average cart abandonment rate near 70%. Both pressures point at the same number, from opposite directions. Getting it wrong in either direction is expensive, which is why a rule of thumb is not good enough.

The three effects a threshold actually has

Before any arithmetic, name the moving parts. A free shipping threshold changes three things simultaneously, and the common guidance models one.

Because the second and third effects pull in opposite directions and both scale with where you set the number, the answer is not a rule of thumb. It is a comparison of contribution margin across candidate thresholds, and the interesting output is not which threshold is best in the abstract but how much conversion lift each one needs to justify itself.

The worked model

Here is a store with round numbers so the arithmetic is checkable. Substitute your own; the shape of the result is what transfers, not the specific figures.

The distribution matters more than the average, so start there. Averages hide the thing you need to see, which is how many orders already clear each candidate threshold.

Baseline contribution margin, holding order count at 1,000: product margin of 45% on $60,220 of revenue is $27,099, plus $7,950 of shipping revenue collected, minus $9,500 of shipping cost. That is $25,549 a month.

Now test three thresholds, holding order count constant so the conversion effect can be isolated afterwards. Uplift assumptions are stated because they are assumptions, and yours will differ: the further a band sits from the threshold, the fewer orders in it will stretch to reach it.

Read the two rows that contradict each other. Average order value rises in all three scenarios, by between 2.8% and 4.9%. Contribution margin falls in all three, by up to 18%. If you were judging this test on AOV, every threshold would look like a win, and the lowest one would look like the biggest win, because it produces the most stretched orders. It is in fact the worst by a wide margin, because 580 orders that were paying $7.95 for shipping stopped paying it.

The output that actually decides it: required conversion lift

None of those thresholds is chosen on that table alone, because the table deliberately held order count constant. The reason to offer free shipping is that it converts more of the traffic you already have, and extra costs at checkout are the leading documented cause of abandonment. So convert each margin gap into the question worth asking: how much more conversion does this threshold need to deliver before it pays for itself?

Divide baseline contribution margin by the new contribution margin per order. The result is the order count needed to get back to level.

That table is the decision, and it reframes the question usefully. A low threshold is easy for customers to reach, so it produces the largest behavioural response, and it needs an implausibly large one to pay for itself. A high threshold is nearly free to offer, and offers little, because a threshold most shoppers cannot reach changes few decisions. The right answer is the middle row, where the lift required is small enough to be achievable and large enough to be worth measuring. On this store that is $75, a shade under AOV plus 25%, which is why the rule of thumb is not useless. It is just not a substitute for checking, because the same rule applied to a store with a wider order distribution or a thinner margin lands in the wrong place.

The formulas, stated plainly

Three calculations do all the work. Keep them somewhere you can re-run them, because the inputs move at least annually.

1. Absolute floor. Shipping cost divided by gross margin percentage. At $9.50 and 45% that is $21.11. Below this, the product margin on the order cannot cover the shipping you just gave away. Treat it as a floor you must clear, never as a candidate threshold: setting the threshold at break-even means you earn nothing extra for the offer.
2. Subsidy leak. Orders already above the threshold, multiplied by your current shipping charge. This is the day-one cost, incurred before any customer behaves differently. In the $50 scenario above it is 580 × $7.95, or $4,611 a month.
3. Required conversion lift. Baseline contribution margin divided by new contribution margin per order, minus one. Compare that against the lift you believe the offer can produce. If the required lift exceeds what is plausible, the threshold is wrong regardless of what it does to AOV.

Two refinements worth applying. Anchor to the mode of your order distribution rather than the mean, because the mean is dragged upward by a small number of large orders and will push your threshold out of reach of where most orders actually land. And round to a clean number: $75 rather than $73, because the threshold has to be memorable enough to change a decision in the cart.

Running it as a test rather than a guess

The required-lift table tells you what to measure, which makes this a genuinely testable change rather than a permanent commitment.

1. Set the success metric before you start, and make it contribution margin per session. Not AOV, not conversion rate, not revenue. Per session rather than per order, because the policy changes order count and a per-order metric will mislead you in exactly the way the table above demonstrates.
2. Run for full purchase cycles, not full weeks. Mixed-basket stores have long tails. If a meaningful share of your revenue comes from customers who deliberate for a fortnight, a two-week test measures the impatient half of your audience.
3. Show progress towards the threshold in the cart. A threshold nobody knows about produces no uplift and pays the full subsidy leak. The progress indicator is not a nice touch, it is the mechanism by which the entire uplift term happens.
4. Disclose the total early. Whatever you decide, the abandonment data is unambiguous that surprise costs late in checkout are the leading cause. The threshold reduces surprise; late-revealed handling fees reintroduce it and cancel the benefit. There is also a compliance dimension to this now, which we covered in all-in pricing laws and checkout.
5. Do not launch it during peak. A structural pricing change in late November has no clean baseline to compare against and no time to reverse. If you did not test it by early November, run the old policy through peak and test in January. The same reasoning applies to the rest of your peak-season change plan, which we set out in the risk-tiered code freeze.

When to recalculate

A threshold is a standing decision based on inputs that all move. Re-run the three formulas when any of these happen, rather than on a calendar.

1. A carrier rate change takes effect. For 2027, do not apply the headline 5.9% to your shipping cost input. Recompute from your own parcel mix, because surcharges are rising faster than base rates and residential and extended-area deliveries carry increases well above the average.
2. Your product margin changes. The floor is shipping cost divided by margin, so a margin compression raises your minimum threshold even when shipping costs are flat.
3. Your order distribution shifts. A successful bundling or upsell programme moves orders above the threshold, which increases the subsidy leak. Growing AOV can make a previously sound threshold unprofitable, which is the counterintuitive one.
4. A competitor changes their offer. Free shipping is now close to an expectation rather than a differentiator, with one industry survey reporting it influences 94% of purchase decisions, so the cost of being the only store charging for it is not symmetric with the gain from being the only one not charging.

If the model says your threshold needs to be higher than your customers can plausibly reach, the threshold is not the lever. The answer is either shipping cost, through packaging, zone skipping or a carrier mix, or order value, through the merchandising work in checkout and AOV optimisation. A threshold cannot fix an order distribution that sits entirely below it, and raising it anyway just moves the abandonment earlier in the funnel. Getting that diagnosis right is most of what conversion work on an e-commerce store consists of.

Frequently asked questions

How do I calculate a free shipping threshold?

Work out three numbers. First the floor: all-in shipping cost divided by gross margin percentage, so a $9.50 shipping cost at 45% margin gives $21.11, below which the order's product margin cannot cover the shipping. Second the subsidy leak: the count of orders that already exceed your candidate threshold, multiplied by the shipping charge you currently collect, which is the cost you incur on day one before any customer changes behaviour. Third the required conversion lift: baseline contribution margin divided by contribution margin per order under the new policy, minus one. Choose the threshold where the lift you can realistically achieve most exceeds the lift you need.

Is the "AOV plus 30%" free shipping rule wrong?

It is incomplete rather than wrong. It models the orders that stretch up to reach the threshold and ignores two larger effects: every order that already qualified now ships free, which forgoes shipping revenue you were collecting, and the conversion rate changes, which is the actual reason to offer free shipping. In the worked example in this article the rule landed close to the right answer on that particular store, but only by coincidence. On a store with thinner margins or a wider order value distribution it lands in the wrong place, and there is no way to tell which case you are in without running the numbers.

Why did my average order value rise but my profit fall after adding free shipping?

Because average order value does not account for the shipping revenue you stopped collecting. When you introduce a threshold, every order that already exceeded it ships free, and if you were previously charging for shipping, that revenue disappears immediately. In the model in this article, a $50 threshold raised average order value by 2.8% while reducing contribution margin by 18%, because 580 of 1,000 monthly orders already qualified and each had been paying a $7.95 shipping charge. Judge the change on contribution margin per session, not on average order value.

What conversion lift does a free shipping threshold need to be worth it?

It depends entirely on where you set it, and lower thresholds need dramatically more. On the store modelled in this article, a $50 threshold required a 22.5% conversion lift to break even, a $75 threshold required 7.3%, and a $100 threshold required 0.4%. The tension is that low thresholds produce the largest behavioural response and need an implausibly large one, while high thresholds are nearly free to offer and change few decisions because most shoppers cannot reach them. The workable threshold is usually in the middle, where the required lift is small enough to achieve and large enough to measure.

How much are shipping costs rising in 2027?

FedEx announced its 2027 general rate increase on 18 September 2026: 5.9% on average across Express, Ground and Freight, effective 4 January 2027. The headline understates the impact for most e-commerce parcels. Reported analysis of the rate tables shows increases of roughly 6.1% to 6.5% on packages from 1 to 20 pounds, with far zones near 6.7%. Surcharges rise faster still: Additional Handling by around 6.8% to 7.6%, the residential Ground Extended Delivery Area surcharge by about 9.1%, and the Oversize charge for zones 5 and 6 from $320 to $345. Recompute your shipping cost input from your own parcel and surcharge mix rather than applying the headline figure.

Should I offer free shipping on everything instead of using a threshold?

Only if your product margin can absorb the shipping cost on your smallest realistic order. Divide your all-in shipping cost by your gross margin percentage: that is the order value at which the product margin exactly covers shipping. If a large share of your orders fall below it, unconditional free shipping loses money on each of them, and the popular fix of building shipping into product prices makes you more expensive on every comparison and search result. A threshold set above the floor keeps the offer while protecting the small orders that cannot fund it.

The takeaway

A free shipping threshold has three effects and the popular rule of thumb models one of them. In the worked example, every candidate threshold raised average order value and two of the three reduced contribution margin, because the orders that already qualified stopped paying for shipping. Calculate three things instead: the floor, which is shipping cost divided by gross margin; the subsidy leak, which is qualifying orders multiplied by your current shipping charge; and the conversion lift each threshold needs to break even. Then choose the threshold where the lift you can realistically achieve most exceeds the lift you need, judge the test on contribution margin per session, and recompute your shipping cost from your own parcel mix rather than from the headline 2027 rate increase.