The two numbers that make the fraction
Revenue is what the period actually took in - total sales before returns and discounts, because a refunded order never paid. Orders is the count of paid orders behind that revenue, not visits, leads or baskets that were abandoned.
Divide the first by the second and the answer is what the average paying customer left in the till each time they checked out. Both numbers have to describe the same window: a month of revenue against a quarter of orders produces a ratio for a period the business never had.
The reverse is the planning half
If every order carries the target basket size, dividing a target revenue by it gives the order count needed to get there. The defaults work this way too: a $20,000 target at a $25 basket needs 800 orders.
That is the number a store can actually act on - more orders of the same size, or the same orders grown bigger. AOV is usually the cheaper lever of the two, because it does not require buying another visit; the order you already paid to acquire simply has to carry more.
The reverse assumes every order carries the target basket, which is the part worth testing. A target set above what the current basket supports is a wish with a division sign in front of it, not a plan.
Whole orders, rounded up
The reverse answer is rounded up to a whole number of orders, because you cannot plan for 833.3 checkouts. The difference between the rounded count and the raw division is small - a reminder that the plan is a target, not a promise.
Rounding up is the safe direction: the plan overshoots the target by less than one order and never falls short of it. A target basket of zero makes the reverse undefined, because there is no figure left to divide the revenue by, and the page prints "not defined" rather than inventing an order count.
What a single average leaves out
One number hides the spread underneath it: a stream of small orders propping the count up while a few large ones hold the average up needs different fixes than a uniform basket.
The page says so in its list, and keeps the reverse read because the ratio alone does not tell you which lever to pull.
The pair worth reading together is AOV, profit per order and the return rate. An average raised by a deeper discount, or by customers who send half of every order back, is not a healthier shop - it is a different one, and the pricing tools in this family are where that difference gets priced.
Where the average meets acquisition cost
An order you already paid to acquire has to keep earning, and that is the budget line an average order value actually competes against. If a visit costs $2 and converts at 4%, the customer costs $50; a $25 basket hands back half of that on the first order, so the second order is where the money is. Lifting the basket by $5 does the same work as cutting acquisition cost by $5 per order, and it does it without buying another visit.
Which is why the useful comparison is rarely this month against a benchmark, and usually against your own cost per order. Mailers, packaging, payment fees and returns all come out of the same basket: a $30 average that costs $12 to serve keeps less than a $27 average that costs $6 to serve, and the page cannot see that difference because it only divides revenue by orders.
Discounts move the number in the direction that flatters it. A 20% code lifts the basket and can still lower the cash you keep, because the extra units went out at a lower margin - so read the average beside gross margin per order, not instead of it. The same asymmetry shows up in free-shipping thresholds: a line set just above the current average pulls the average up by construction, and the orders that reach it usually get there with a cheap add-on rather than a profitable one.
Keep the window fixed when you compare. Two months of different length, or a month carrying a promotion, are not a trend - the ratio moves with the calendar. Measure at the same point in the order lifecycle too: a basket counted before returns is a different number from the same basket counted after them, because a refunded order never paid, and the reverse plan is only as good as the revenue figure fed into it.
Danfio