If your average order value is £45 and you set free shipping at £50, every customer who adds a £5 item to qualify costs you £2.75. Your average order value rises 11%. Your profit falls on every one of those orders.
The threshold formula nobody publishes
Free shipping thresholds are the most common AOV tactic and the most commonly set by guesswork — usually "a bit above our average."
There's an actual break-even, and it's simple:
The rule Minimum gap above your AOV = shipping cost ÷ gross margin. Below that gap, the extra spend doesn't cover the shipping you're giving away.
With £5 shipping and a 45% gross margin, you need roughly £11 of additional spend just to break even. Which means a threshold set anywhere under about £56 on a £45 average is destroying margin on the orders that just clear it.
| Shipping cost | 30% margin | 45% margin | 60% margin |
|---|---|---|---|
| £4 | £13.33 | £8.89 | £6.67 |
| £5 | £16.67 | £11.11 | £8.33 |
| £8 | £26.67 | £17.78 | £13.33 |
| £12 | £40.00 | £26.67 | £20.00 |
Worked through on a £45 AOV with £5 shipping and 45% margin:
- Threshold £50: customer adds £5, contributes £2.25 margin, you absorb £5 → −£2.75
- Threshold £55: adds £10, contributes £4.50 → −£0.50
- Threshold £60: adds £15, contributes £6.75 → +£1.75
- Threshold £70: adds £25, contributes £11.25 → +£6.25
Note that low-margin retailers need a much larger gap, which is counterintuitive — the businesses least able to absorb shipping are the ones most tempted to set a tight, achievable threshold.
AOV on its own is a trap
The deeper problem with this whole category, and the reason a lot of AOV work quietly destroys value.
Average order value can rise while profit falls. Every mechanism below does it:
- A threshold set too low — as above.
- Discounting to hit the threshold. "Spend £50, get 10% off" on a £45 average means you paid roughly £5 to gain £5.
- Reduced conversion. Upsells and interruptions before payment cost you some completed purchases. A 6% AOV gain against a 4% conversion loss is close to a wash before you count anything else.
- Returns. Items added to qualify, or bundled products nobody specifically wanted, come back at higher rates than deliberate purchases.
- Bundle cannibalisation. Discounted bundles capture people who'd have bought the items separately at full price.
The metric that actually matters is contribution profit per visitor. It accounts for AOV, conversion rate, discount, shipping and margin simultaneously, which is exactly the set of things that trade against each other here. Any tactic that raises AOV while lowering that number is a loss dressed as a win — and it's the kind of thing that only surfaces if you set the goal properly in the first place.
Returns are the omitted cost
Worth its own section because it's almost never in the calculation and can invert the result entirely.
A returned item costs you outbound shipping, return shipping, processing labour, and frequently some resale value. In apparel, where return rates are structurally high, an item added purely to clear a shipping threshold is a strong candidate to come back.
So a tactic can lift checkout-value AOV by 12% and lift net-revenue-after-returns by considerably less, or not at all. Measure to net revenue after the return window closes, not to the value at the point of purchase.
This is also an argument for the tactics that add relevant items rather than any items. A genuinely complementary product is kept; a filler item bought to save £5 of postage is returned.
Printify
Print-on-demand platform to design and sell custom products with no upfront inventory.
Best for: Creators and stores selling custom merch with print-on-demand, no inventory
Placement decides the risk
Upsell, cross-sell and order bump get discussed as different offers. The more useful distinction is where they appear, because that determines what they can cost you.
| Placement | Risk to the original sale | Best used for |
|---|---|---|
| Product page | Moderate — can create indecision | Genuine upgrades and variants |
| Cart | Moderate — adds a decision before checkout | Threshold prompts, obvious accessories |
| Checkout (order bump) | Higher — any friction here is expensive | One low-price, one-click item only |
| Post-purchase | None — payment already taken | Anything. Free upside. |
| Post-delivery email | None | Consumables, replenishment, companions |
The post-purchase slot is the underused one. An offer presented after payment has been captured cannot cost you the original sale — the worst outcome is that nobody takes it. Every other placement trades conversion risk for AOV, and post-purchase doesn't. If you implement one thing from this article, make it that.
Post-delivery follow-up belongs in the same category and runs on infrastructure you probably already have — the flows covered in core email automation extend naturally to replenishment and companion offers timed to when the first purchase is actually being used.
Which tactic fits your catalogue
Most guides hand every retailer the same ten tactics. What actually works depends on the shape of what you sell.
Few SKUs, high price: upsell to a better version, extended warranty or service. Bundling has little to work with, and thresholds matter less when a single item already clears any sensible one.
Many SKUs, low price: thresholds and bundles do the heavy lifting. This is where the break-even maths above matters most, because margins are usually thinner and the gap needs to be wider.
Consumables: quantity breaks and subscription are the highest-return mechanics, and they improve lifetime value rather than just this order. Selling three months instead of one raises AOV and reduces the chance the customer buys elsewhere next time.
Durables with accessories: cross-sell is natural and largely non-cannibalising, since the accessory purchase wasn't happening otherwise. This is the cleanest AOV win available to anyone who has it.
Gifting-heavy categories: gift wrap, cards, multi-recipient options. High margin, genuinely wanted, rarely returned — and particularly relevant heading into peak, as covered in the 90-day BFCM timeline.
Bundles that work and bundles that don't
Bundling is the tactic most likely to look successful and be unprofitable.
What works: pairing a high-margin item with a lower-margin one, so the blended margin holds while the customer perceives a saving. Or solving a completeness problem — the set someone would otherwise have to assemble themselves, where the value is convenience rather than discount.
What doesn't: discounting a bundle of things people were already buying together at full price. That converts full-margin revenue into discounted revenue with no volume gain — a genuine loss recorded as an AOV win.
The test: would these items have been bought together anyway? If yes, bundling them at a discount costs you money. If no, the bundle is creating the additional sale and the discount is buying something real.
Presentation matters as much as construction. A bundle needs to make the saving obvious and the components clear — which is product page work rather than merchandising, and covered in product page optimisation.
Testing this properly
An honest caveat, because AOV changes are harder to test than they look.
AOV is a continuous measure with high variance — a handful of unusually large orders can swing the average considerably. That makes it noisier than conversion rate and generally requires more data to read reliably, not less. The sample-size discipline in post-click testing applies here with an extra margin.
Two practical consequences. Use median order value alongside the mean, since the median is far less distorted by outliers and often tells a truer story about what typical customers did. And measure the composite — contribution per visitor — rather than AOV in isolation, so a conversion loss can't hide behind an average gain.
Where volume doesn't support a clean test, the threshold maths above is at least deterministic: you can compute whether a given threshold is above break-even without any experiment at all.
The order to work in
- Calculate your break-even threshold gap. Shipping cost ÷ gross margin. If your current threshold is below it, that's a same-day fix.
- Look at your order value distribution, not just the average. Where do orders actually cluster? That constrains where a threshold can sensibly sit.
- Add a post-purchase offer. No conversion risk, so it's the only genuinely free upside in this list.
- Show progress to the threshold in the cart. "£9 away from free delivery" outperforms stating the threshold, because it converts an abstract rule into a specific small action.
- Add cross-sells only where genuinely complementary. Relevance protects you against returns.
- Check returns by tactic after a full return window. This is where the truth is.
- Only then consider bundles, and apply the would-they-have-bought-it-anyway test to each.
Step four is worth its own note. The progress indicator works because a shopper £9 short of free delivery is being asked one small concrete question rather than being shown a policy. It's also one of the few AOV mechanics with no downside — it doesn't discount, doesn't interrupt, and doesn't add a decision to the checkout itself, which puts it alongside the fundamentals in conversion-focused design.
One timing note: heading into a season where value-seeking is unusually widespread, threshold and bundle mechanics land differently — shoppers actively hunting savings respond well to a clear route to free delivery, and badly to anything that reads as a trick. The behavioural picture is in what early signals say about holiday spend.
If margins are tight enough that the break-even calculation rules out a workable threshold entirely, the constraint is pricing and unit economics rather than merchandising — and that's a different conversation from an AOV tactic, one where an outside e-commerce partner looking at the whole picture usually helps more than another cart widget.
The short version
Your minimum free shipping threshold gap is shipping cost divided by gross margin — around £11 above AOV at £5 shipping and 45% margin, which means a threshold set £5 above your average loses money on every order that just clears it. Judge every tactic on contribution profit per visitor rather than AOV, because AOV rises happily while profit falls through discounting, lost conversion and returns. Measure after the return window, since filler items come back. Use post-purchase offers first, as they're the only placement that can't cost you the original sale. And before bundling anything, ask whether those items would have been bought together anyway — if so, the discount is pure loss recorded as a win.
Average order value climbing while profit stays flat?
We rebuild merchandising and offers around contribution per visitor rather than headline averages.
Explore Performance Marketing →Frequently asked questions
Where should you set a free shipping threshold?
Far enough above your average order value that the margin on the extra spend covers the shipping you give away. The break-even gap is your shipping cost divided by your gross margin percentage. With a £5 shipping cost and a 45% gross margin, you need roughly £11 of additional spend just to break even — so a threshold set only £5 above your average order value actively loses money on every order that just clears it, even though the headline average rises.
Is increasing average order value always good?
No, and treating it as a standalone goal is the most common mistake in this area. Average order value can rise while profit falls, because the tactics that raise it frequently reduce conversion rate, absorb shipping costs, apply discounts, or generate purchases people did not really want and later return. The metric that actually matters is contribution profit per visitor, which accounts for all of those effects simultaneously. Any tactic should be judged on that rather than on the average alone.
What is the difference between an upsell, a cross-sell and an order bump?
An upsell offers a better or larger version of what someone is already buying. A cross-sell offers a complementary product alongside it. An order bump is a small, cheap, one-click addition presented at the point of checkout. They differ mainly in where they appear and what they risk: offers placed before payment can distract from the purchase and reduce conversion, whereas an offer presented after payment has been taken cannot cost you the original sale at all.
Why do bundles sometimes reduce profit?
Because they are often discounted to encourage the larger purchase, and the discount can exceed the margin gained on the extra items. A bundle also risks cannibalising sales of items customers would have bought separately at full price, which converts full-margin revenue into discounted revenue without adding volume. Bundles work best when they combine a high-margin item with a lower-margin one, or when they genuinely solve a completeness problem the customer had, rather than when they simply reduce the unit price.
How do returns affect average order value strategies?
Substantially, and they are usually omitted from the calculation. Items added to reach a shipping threshold, or bundled products the customer did not specifically want, are returned at higher rates than deliberate purchases. In categories like apparel this can erase the gain entirely, since a returned item costs you outbound shipping, return shipping, processing and often the resale value. Any average order value initiative should be measured on net revenue after returns rather than on the value at the point of checkout.
