I have run the AOV play from both chairs. I ran my own brand for 14 years and I now sit inside the numbers of more than a dozen other brands. The pattern is the same everywhere. Someone reads a list of 17 ways to increase average order value, installs a cart upsell app, sets a free shipping threshold at a round number, and three months later the dashboard says AOV is up 11 percent and the P&L says nothing changed. This is how to do it so the P&L notices.
Average order value is the wrong goal, and that is the whole point
AOV is a ratio, not an outcome. Revenue divided by order count. You can push it up by selling more things per order, by selling more expensive things, or by selling fewer cheap orders. Two of those make you money. One of them, the one most apps are built around, mostly moves money from one line of the P&L to another.
The goal underneath AOV is profit per order, and underneath that, profit per customer. Keep that in view and half of the standard advice falls away on its own. A bundle that takes 20 percent off three products raises AOV and can lower the profit on every order that takes it. A spend-$50-get-a-gift offer raises AOV and hands your margin to customers who would have spent $50 anyway. These are not mistakes in execution. They are the wrong goal, executed well.
The question I ask before any order-value work is the same one I ask before any test: which number are we trying to move? More units in the box, a pricier unit in the box, or more dollars left after the box ships? Say it out loud. The first two are means. The third is the end, and it changes which tactics are allowed.
Where the standard tactics actually land
The 17-way lists are not wrong about what moves the AOV number. They are silent about what each tactic does to margin and to the next order, which is the only thing that matters. Here is the same list with the two missing columns.
| Tactic | Effect on profit per order | Effect on the next order |
|---|---|---|
| Discount bundle (3 for 20% off) | Down, unless the margin on the third unit covers the discount on all three | Down, the customer stocked up |
| Free shipping threshold | Up, if the threshold sits above your current AOV and shipping costs less than the added margin | Neutral |
| Spend X, get a free gift | Down on orders that would have cleared X anyway | Neutral |
| Complementary item in the cart at full price | Up on every order that takes it | Neutral |
| Multipack or set at full price per unit | Up | Down, the customer stocked up |
| Raising prices | Up on every order | Depends on whether the product earns it |
Read the middle column first. Two of the six lower profit per order on the orders you already had. That is the trade nobody writes down: the tactic raises the average by giving something away to the people who were already going to buy. Then read the right column. Anything that gets a customer to stock up today pulls forward an order you would have gotten next month, which looks like growth for one quarter and like churn the next.
That right column is the one I care most about, because the mistake it describes is the one I see most. A brand finds a product or a customer group that converts, and every tactic gets pointed at squeezing more out of that one slice. AOV goes up. Repeat rate goes down. Total pie, unchanged or smaller. The job is to grow the whole pie, new customers and returning ones together, and an AOV tactic that eats the next order is working against the brand while the dashboard says it is working for it.
The three levers that raise profit per order
What you sell and at what price
Assortment and price set your AOV before any app touches it, and they are the lever most teams never pull because they live in merchandising, not marketing. If your hero product is $34 and everything else is cheaper, no upsell widget gets you to a $70 average. A $60 version of the hero product does, and so does a two-unit set priced at two units.
The operator move is to price the second unit, not discount it. A set of two at $68 instead of two singles at $34 raises AOV with zero margin given away, and the customers who want two were going to buy two. If you need a reason for the set to exist, make it the packaging or the pairing, not the price.
The same logic says raise prices before you add bundles. A price increase lands on every order, needs no app, and tells you quickly whether the product earns it. Brands add one discounting mechanic after another and leave the price where it was. That is backwards.
The complementary item, in the cart, at full price
The most reliable way to add a unit to an order is to offer the thing that goes with the thing, at the moment the customer is already saying yes, without a discount. The refill with the device. The case with the frames. The second flavor next to the first. This is different from a bundle, because nothing is marked down, and different from a generic upsell widget, because the item is chosen for relevance rather than pulled from the bestseller list.
Most cart recommendation apps get this wrong in the same two ways. They show six items when one or two would convert better, and they show whatever sells most instead of what pairs with what is already in the cart. A customer buying a $120 jacket does not need to see your three top sellers. They need to see the $28 item that completes the jacket, and that is the one that gets added without a second thought.
The rules I use: the add-on is complementary, not a substitute; it is priced below the item it attaches to, so it reads as a yes rather than a decision; it is full price; and there is one of it, two at most. Measure attach rate, the share of orders that take the add-on, by hero product. A low attach rate means the pairing is wrong, not the widget. This lever does what the discount bundle pretends to do, more units per order, and keeps every dollar of margin while doing it.
The free shipping threshold, set from your numbers
Of the tactics every brand already runs, the free shipping threshold is the one most often set wrong, and the one that pays back most when it is set right. The right threshold is not a round number and it is not whatever a competitor uses. It comes from three figures you already have: your current AOV, your gross margin, and your average outbound shipping cost.
Set the threshold above current AOV by roughly the price of your most common add-on item, so a meaningful share of carts has to add one item to clear it. Then check the math on the added item: its gross margin has to exceed the shipping cost you are now eating. If a customer adds a $12 item at 60 percent margin to clear free shipping, you gained $7.20 of margin and gave up $8 of shipping revenue. That threshold loses money. Move it, or change what sits at the add-on price point.
After that, leave it alone for a full quarter and read profit per order, not AOV. A threshold that moved AOV by 8 percent and margin per order by 1 percent did its job. One that moved AOV by 8 percent and margin per order by nothing was a free shipping giveaway with extra steps.
How to measure it so you do not fool yourself
Measure contribution margin per order, before and after, on the same customer mix. Contribution margin is revenue minus product cost, shipping, payment fees and the cost of the discount. If you only watch AOV, every discounting tactic will look like a win, because discounts raise the average and hide in the cost lines.
Split new customers from returning customers before you read anything. A tactic can raise AOV on returning customers and crater it on new ones, or the reverse, and the blended number will tell you nothing. Then give it a full quarter. Order value tactics change what people buy next, and the first week’s dashboard cannot show that.
The whole pie is one line of arithmetic: Customers, times Orders per customer, times Profit per order, equals Profit. AOV, after costs, lives in Profit per order, and a tactic that raises it by cutting Orders per customer has grown nothing.
And write the goal down before you start. More units per order, a pricier unit per order, or more profit per order. If the tactic you are about to run does not move the third one, it is a dashboard tactic, and you already have enough of those.
Questions people ask
How do you calculate average order value?
Divide total revenue by the number of orders for the same period. $500,000 in revenue across 8,000 orders is a $62.50 average order value. Use net revenue after discounts and returns, or the number flatters you. Calculate it separately for new and returning customers, because the two groups behave differently and the blended figure hides which one moved.
What is a good average order value?
There is no good number, only a good number for your margin structure. A $40 AOV at 70 percent gross margin beats a $70 AOV at 30 percent. Benchmarks by industry exist and are mostly useless, because they average brands with different products, shipping costs and price points. Compare your AOV with your own last four quarters, split by new and returning customers, and judge it by profit per order.
Is a high average order value always good?
No. AOV rises when you discount bundles, when customers stock up, and when you lose your smaller orders. The first costs margin, the second borrows from next quarter, and the third is shrinking revenue dressed up as a better ratio. A high AOV is good when profit per order rose with it and order frequency did not fall. Check all three before you celebrate.
Does a free shipping threshold increase average order value?
Yes, when it sits above your current AOV by roughly the price of a typical add-on item and the margin on that item exceeds the shipping cost you absorb. Set at or below current AOV, it is a shipping giveaway. Set too high, nobody reaches it and it stops mattering. Set it from your own margin and shipping figures, run it for a quarter, and read contribution margin per order, not AOV.
Upsell or cross-sell, which raises order value more?
A complementary cross-sell at full price raises order value more reliably than an upsell. The upsell asks the customer to swap to a pricier version of what they already chose, which reopens the decision. The cross-sell adds the thing that goes with it, priced below the main item, which reads as a yes. Either one offered with a discount raises the AOV number and lowers the profit on the order, so hold the price.
Klaviyo email marketing for DTC brands
Flows, campaigns, design and deliverability, run as one program by senior people.
See how we run itIf you want one thing to do this week, pull contribution margin per order for the last two quarters, split by new and returning customers, and look at it next to your AOV. If AOV went up and margin per order did not, you already know which tactics to turn off. The rest of this argument, the goal-first version of testing, is in the CRO piece. The other half of the pie, the next order rather than this one, is what our email team works on. If you would rather we look at your numbers with you, start here.
