Increase Average Order Value Ecommerce: Levers vs Discounts

What actually increases average order value in ecommerce?

Teams that want to increase average order value ecommerce-wide usually reach for the same three levers: bundling, a free-shipping threshold, and a post-purchase offer. Only one of them reliably raises both order value and margin at the same time. The other two often just move a discount around and call it growth — a shipping threshold priced wrong, or a post-purchase offer that undercuts the item it’s attached to, can lift the number on the order confirmation screen while lowering the number that pays for the warehouse.

This matters most if you’re already running one or two of these levers without having checked which one is actually contributing margin. At $3M–$30M in revenue on Shopify Plus or a comparable subscription platform, you likely have all three switched on somewhere in the funnel already. The question isn’t whether to add a fourth lever. It’s whether the three you have are pulling in the same direction, or whether one of them is quietly financing the other two.

Does bundling increase average order value, or just shift revenue forward?

Bundling increases average order value when the bundle price sits above what the customer would have spent on the anchor item alone, and it protects margin when the attached item carries a similar or better margin rate than the anchor. It turns into a disguised discount the moment the bundle price gets set by subtracting a round-looking number from the sum of the parts, instead of working up from the cost of each component.

The failure mode is ordinary and easy to miss. A merchandiser picks a round discount because it looks right on the product page, without checking what that number does to blended margin once both items’ cost of goods is accounted for. At low order volumes this is a rounding error nobody notices. At scale, that unreviewed discount rate compounds against thousands of transactions a month and shows up as a quarter where AOV is up and gross profit dollars are flat.

How do you price a bundle so it doesn’t erode margin?

Here’s an illustrative example, not a benchmark: a core product sells for $40 at a 60% gross margin (cost $16), and an attachment item sells for $15 at a 40% margin (cost $9). Sold separately, combined revenue is $55 against $25 of cost, a 54.5% blended margin. Price the bundle at $48 — a $7 discount off the combined price, roughly 13% — and that same $25 of cost now sits against $48 of revenue, a 47.9% blended margin.

Order value went up, assuming the attachment item wouldn’t have sold on its own: the order moved from a $40 single-item purchase to a $48 bundle. But margin rate dropped by more than six points on that order. Whether the trade is worth it turns on one number you need before setting the bundle price: the attachment item’s standalone attach rate. If it barely sells on its own, the bundle is close to pure incremental margin. If it already sells at a reasonable clip alone, the bundle is cannibalising a full-margin sale and replacing it with a discounted one.

What’s the right free-shipping threshold for average order value?

There isn’t a single right threshold, but there is a right way to set one: start from your current AOV and your per-order shipping cost, then set the threshold high enough that the marginal revenue it pulls in covers the shipping cost you’re about to absorb. Set it below that line and you’re subsidising orders that were never going to grow.

Illustrative maths, not a benchmark figure: say your average order sits at $54, your outbound shipping cost per order runs $9, and your gross margin on incremental goods is 58%. Under a paid-shipping model, the customer covers that $9, and your gross profit on the order is 58% of $54, or $31.32. Set the free-shipping threshold at $65, and if a customer adds $11 of goods to clear it, you now absorb the $9 shipping cost yourself. Gross profit becomes 58% of $65 minus $9, or $28.70 — lower than before, even though the order itself got bigger.

The threshold only pays for itself once the margin on the added goods exceeds the shipping cost you’re giving up. In this threshold example, the $11 of added goods contributed $6.38 in margin (58% of $11) against a $9 shipping cost — a $2.62 shortfall per order at that threshold. Push it to $70 or $75 and the added-margin side of the equation grows faster than the shipping side, because shipping cost per order doesn’t move with the threshold. The break-even point is specific to your own AOV, margin and carrier rate, not a number you can borrow from a case study.

What breaks when the free-shipping threshold is set wrong?

A threshold set too low costs you on every order that would have converted anyway. If your typical cart already clears $50 and the free-shipping trigger sits at $45, you’re giving away shipping on orders that needed no incentive to complete — pure margin loss, no behaviour change to show for it.

A threshold set too high does the opposite: almost nobody reaches it, so it neither changes behaviour nor costs you anything, and the free-shipping banner becomes background noise on the page. The tell is in your cart data: if carts stall at or just under the threshold with nothing added afterward, the number is too high for what you’re asking customers to add. If carts clear it easily on a single item, it’s too low.

There’s a volume problem underneath this too. At low order counts, you can eyeball the issue by reading a week of cart sessions. Past a few hundred orders a day, the threshold needs checking against a rolling AOV trend, not a static number set once at launch and left through two years of price increases. A threshold set against last year’s AOV gets quietly cheaper for the customer to hit every time you raise prices, which raises your shipping subsidy without anyone deciding it should.

Do post-purchase offers increase average order value, or just discount the order?

A post-purchase offer — the one-click add shown on the order confirmation or thank-you page — increases average order value only when it’s priced independently of the checkout discount the customer just used. It becomes a disguised discount when the offer repeats the same 15% off the customer already had in their cart, presented a second time after they’ve committed to buy. That’s not incremental revenue; it’s the same margin given away twice, once on the original order and again on the confirmation page.

The honest test is to run the offer with no discount at all for a fixed period and compare acceptance rate against the discounted version. If acceptance barely moves without the discount, the discount was never earning its keep — you were paying margin for an add-on the customer would have taken anyway.

For consumable or repeat-purchase categories, timing matters as much as pricing. An offer pitched at the moment of purchase competes with the customer’s actual need: if they won’t run out of the product for another six weeks, no discount makes a second unit useful today. Pointerflow’s replenishment timing calculator estimates the date a customer is likely to reorder from their purchase history, which is a better trigger for a follow-up offer than the checkout moment itself — a lever most teams place at exactly the wrong point in the cycle.

Which AOV lever is really just a discount in disguise?

Read the table row by row rather than as a ranking. Each lever’s middle column is the discount it turns into when priced wrong, and the fix is specific to that lever, not a blanket rule that applies to all three.

How does order volume change which AOV lever you pull first?

At under a few hundred orders a month, bundling is the cheapest lever to test because you can hand-price two or three bundles and watch the effect directly in weekly revenue. A free-shipping threshold needs volume to read cleanly: at low order counts, a handful of large orders from repeat customers who’d have bought anyway will swamp the signal from customers actually responding to the threshold.

Past that volume, the constraint flips. Hand-priced bundles stop scaling once you’re running dozens of them across a catalogue where costs move independently — a bundle priced correctly in January can be quietly underwater by March if one component’s cost rose and nobody re-ran the numbers. This is the point where a free-shipping threshold, reviewed against a rolling AOV figure rather than set once, becomes the more reliable lever, because it’s one number to keep current instead of dozens of bundle prices.

Post-purchase offers scale differently again. The offer mechanism itself doesn’t care about order volume, but the discount decisions behind it do. At high volume, a poorly tuned post-purchase discount is the fastest way to erode margin without anyone noticing, because it’s live on every order and rarely gets the scrutiny a headline promotion would.

What should you measure before you pick an AOV lever?

Before running any of these three, you need your current AOV, your margin rate by product or category — not blended across the whole catalogue — and your shipping cost per order. Without the margin rate broken out, you can’t tell whether a lever that raised AOV also raised gross profit, or just moved revenue between two SKUs at different margins.

You also need a read on attach rate: how often the item you’d bundle, threshold-incentivise, or upsell post-purchase already sells on its own. A lever built around an item with a high standalone attach rate mostly cannibalises sales you already had. One built around an item with a low standalone attach rate is closer to genuinely incremental.

If any of these numbers isn’t tracked yet, that’s the actual first step, not a benchmark AOV figure to chase. A target borrowed from an industry report tells you little about your own margin structure — your category, cost base and typical basket size all differ from whatever backed that number. The transferable part is the method for setting your own threshold and bundle price, not the figure itself.

How do you increase average order value on Shopify specifically?

The three levers work the same way on Shopify Plus as anywhere else; what changes is where the settings live. Bundling is usually built through a bundle or “frequently bought together” app, or through native product bundle functionality where the platform supports it, and the same bundle margin math applies whether the discount is coded as a percentage-off or a fixed bundle price. The free-shipping threshold sits in shipping settings or a promotions app, and it’s worth checking whether it applies storefront-wide or by product tag, because a single global threshold treats a heavy, expensive-to-ship item the same as a small one with a much thinner shipping cost.

Post-purchase offers on Shopify typically run through a checkout extension or a dedicated post-purchase app, since native checkout doesn’t include an upsell step by default. Whichever app you use, check whether the discount applied there is tracked separately from your other discount codes in reporting. If it isn’t, the margin cost of that lever gets buried inside a general discounts line, and you lose the ability to tell whether it’s paying for itself.

How do you A/B test a lever to increase average order value ecommerce-wide without corrupting the read?

Change one lever at a time against a genuine holdout, and size the test window to cover at least one full purchase cycle for repeat-heavy categories. A bundle or threshold change tested for three days catches whichever customers happened to shop that week, not a representative read of how the lever performs across a normal mix of new and returning traffic.

Track margin and attach rate alongside AOV in the same test, not as a follow-up analysis afterward. A test that only reports the AOV delta can show a lever “winning” while the margin data, checked separately weeks later, shows it was funded by a discount that outran the incremental revenue. Pull both numbers from the same test window so the trade-off is visible before you decide to roll the lever out permanently.

Watch for concurrent promotions running in the same window. A sitewide sale launched mid-test will move AOV on its own, independent of the lever being tested, and crediting the lever for a lift that was really the promotion means you keep running something that isn’t doing the work you think it is. If a promotion is unavoidable during the test period, extend the window past it rather than reading results while it’s still live.

At low order volumes, a single unusually large order can swing a short test’s average enough to look like a result. Set a minimum transaction count for the test before you start — a threshold you decide in advance, not one you pick after seeing which number makes the lever look good — and don’t call a result until the test clears it. Two levers changed in the same week is the most common way a test gets corrupted: if a bundle price and a free-shipping threshold both move in the same seven-day window, neither change gets a clean read, and the next month’s numbers won’t tell you which one to keep.

Who shouldn’t make AOV the first thing they fix?

If your margin rate by product isn’t tracked, or your shipping cost per order isn’t known separately from a blended fulfilment rate, running any of these three levers means optimising against a number you can’t verify. Fix the measurement first. Chasing a higher AOV number without a margin figure behind it can make a quarter look better while gross profit dollars stay flat or fall, and the dashboard moves in the wrong direction to make anyone notice.

It’s also not the first lever for a brand still working out repeat purchase rate or churn. AOV describes a single transaction; a customer who buys once at a high order value is worth less over time than one who buys at a modest order value three times over. If the immediate problem is getting customers to come back at all, that’s a retention problem before it’s an AOV problem, and the two need different fixes even though they sit in the same part of the funnel.

Getting the pricing right on a bundle, setting a threshold that pays for itself, and timing a post-purchase offer against real purchase behaviour instead of the checkout moment are all the same underlying problem: connecting what happens after a customer decides to buy to what it actually costs you to have them buy more. That’s the problem Pointerflow’s post-purchase and AOV work is built around.