Average order value playbook: the complete guide

One store wide average is almost useless for making decisions, because it blends groups that behave nothing alike. Break it apart before you plan anything.

Split the number four ways. By channel, because a visitor from a search for a specific product behaves differently from someone who arrived through a social ad. By new versus returning customer, because returning buyers usually spend more and any change in traffic mix will move the blended average without anything real happening. By device, because phone baskets and desktop baskets differ. And by category, because a store selling both a low priced consumable and a high priced piece of equipment has two businesses in one average.

Look at the shape of the distribution as well as the average. Export your orders and count how many fall into each band of order value. Most stores find a large group clustered close together with a thin tail of much larger orders, and that tail can drag the average well above where a typical order sits. The median tells you what a normal order looks like. The average tells you what the business earns. You need both, and a decision made on the average alone will usually be set too high.

Count items per order alongside the money. It is the cleanest signal of whether bundling and add on work is having any effect, because it cannot be flattered by a price rise or by a change in the mix of what people buy.

Then fix your baseline in writing. Order value and items per order, for the last twelve months, split the four ways above, with the median beside the average. Every move in this playbook gets judged against that baseline, and you cannot rebuild it later once the store has changed.

Rebuild the same split every quarter. Order value shifts slowly and quietly, and the only way to notice one category drifting downwards is to look at the same breakdown on the same schedule.

Your own orders already say which move will work. Before deciding between bundles, thresholds, add ons or larger sizes, spend an afternoon reading what people actually put in the same basket.

Export a year of order lines and answer four questions. Which pairs of products appear together most often, and how often relative to how often each sells on its own. How many orders contain exactly one item. Which products are almost always the only thing in the order. And which products, when they appear, tend to bring a larger basket with them.

Those answers point at different work. If most orders contain exactly one item and there is no obvious pairing, you have an add on and price ladder problem: nothing on the site suggests a natural second item. If clear pairs already exist, you have a bundling opportunity that requires no invention, only presentation. If one product drags the whole basket up, it deserves better placement and more traffic. If a product is always alone and cheap, it may be costing you money once shipping is counted.

Read the near misses too. Orders that sit slightly below a shipping threshold, carts abandoned with two items where one was removed, and products frequently viewed in the same session but rarely bought together. Sessions where someone viewed a case, a filter or a refill and did not buy it tell you the pairing is real but the prompt is missing.

One warning. Products bought together because they sit next to each other in your navigation are not a genuine pairing, they are a layout artefact. The test is whether the pair makes sense to a customer explaining it out loud. A blender and a spare jug make sense. A blender and a tea towel do not, no matter what the report says.

The average order value number can rise while the money you keep falls. That happens whenever the lift is bought with a discount, an extra parcel, a heavier box or a product with thinner margin. So the working number for this playbook is not order value. It is contribution per order: what is left after the cost of goods, the payment fee, the packaging and the shipping you actually pay.

Build that figure once, properly. Take a typical order and subtract the landed cost of the goods, the payment processing fee, the pick and pack cost, the packaging, the outbound shipping cost you pay the carrier, and a realistic allowance for returns in that category. What is left is what an order is worth to the business. Do this for your two or three main product groups, because they will differ more than you expect.

Now every idea gets tested against it. A bundle sold with money off has to raise contribution after the discount. A free shipping threshold has to raise contribution after paying for the parcels of every order that would have qualified anyway. A larger size has to earn more per order than two of the small size did, once the heavier parcel is paid for.

Worked example, with your own numbers substituted. Say a typical order is sixty dollars and leaves twenty four dollars of contribution. A bundle raises the order to eighty five dollars but includes ten dollars off and an item that costs you eighteen. The order value has risen by a quarter, and contribution has moved by a little over five dollars. That is still a win. Change the discount to twenty dollars and the same bundle takes money out of the business while looking like a success in every report.

This is why the reporting chapter at the end of this guide matters. If you only watch the average, you will keep the moves that flatter it.

There are six ways to raise order value and most stores can only do two of them well. Choosing badly wastes months, so match the move to what you sell rather than working through the list.

If you sell consumables that run out, quantity and subscription are the strongest moves. A three pack with clear value per unit, or a subscription that saves a little and removes the reordering, raises the order and the repeat rate at once. If you sell equipment or durables, accessories and warranties are the move: the case, the spare part, the refill, the extended cover. If you sell fashion or homewares, sets and coordinated pieces work, because the buyer is already thinking about how things go together. If you sell one product with no variants, quantity tiers and gifting are usually the only honest options.

If you sell many low priced items, a free shipping threshold does more than anything else, because shipping is a large share of the order and the buyer feels it. If your average order is already high and shipping is a small share, a threshold changes very little and you are better off on merchandising and price ladders.

Write down the two moves you are choosing and the reason, then ignore the rest for a quarter. Stores that try all six at once end up with a product page carrying a bundle offer, a quantity discount, an add on prompt, a threshold bar and a subscription toggle. That page converts worse than the plain version, and the order value work has cost conversion, which is the more valuable lever.

The rest of this playbook covers the shared craft behind whichever two you pick: the price ladder, where to put the prompt, how the shipping rules help, and what to do about orders too small to be worth packing.

A price ladder raises order value without asking anyone to buy a second product. You give the same buyer a bigger version of the decision they were already making, and you make the value obvious enough that the middle or larger option is the comfortable choice.

The ladder has three parts. An entry option that exists so the range has a floor, a main option that most people should choose, and a larger option that makes the main one look reasonable. Sizes, multipacks, durations and kit levels all work as rungs. The point is not to trick anyone. It is that a single price with nothing to compare it to gives the buyer no way to judge whether it is fair.

Show the per unit value at every rung. Price per hundred millilitres, price per serving, price per month, price per pair. When the larger size is genuinely better value, saying so plainly moves a real share of buyers up a rung, and it is the most honest form of order value work there is, because the customer gets more for less per unit.

Gap the rungs so the step up is easy. If the two pack is nearly the price of two singles, nobody moves. If it saves enough to notice but not enough to hurt your margin, plenty do. Look at the actual gap in money as well as the ratio, because a small step in money is easy to accept regardless of what the maths says.

Default to the option you want chosen. Whichever rung is pre selected when the page loads gets chosen more often than the others, so select the main option rather than the cheapest, and make sure the price shown at the top of the page matches that selection so nothing changes under the buyer.

And keep the ladder to three rungs where you can. Four or more turns a simple decision into research, and research is where orders get postponed.

A relevant suggestion in the right place converts. The same suggestion in the wrong place is ignored, and a wall of suggestions anywhere costs you attention you needed for the first item. There are only four places worth using, and each one has a different job.

On the product page, below the buying block, show what genuinely goes with this item. Two or three products, chosen from your basket analysis rather than from an algorithm nobody has checked, with a short reason for each. Complete the set, or add the refill, or the case that fits this model. Never show competing versions of the product the buyer is already looking at, because that restarts the decision rather than extending it.

In the cart or the drawer, show one thing. This is the highest converting placement in most stores, because the buyer has committed and is thinking about whether they have everything. Choose something cheap relative to the basket, obviously related, and addable in one tap without leaving the cart.

On the collection page, merchandise the sets and multipacks so people meet them before they have anchored on the single item price. A shopper who sees the two pack first evaluates the single as the small option rather than the normal one.

After payment, on the confirmation page, you can make a bolder offer because nothing is at risk. This is the right home for the add on that is too distracting to show before payment.

What all four have in common is restraint. One idea per placement, chosen from real basket data, described in the customer's words. Recommendation widgets that fill a page with loosely related products train buyers to scroll past the whole area, which costs you the placement permanently.

Review what each placement is showing once a quarter. Products go out of stock, get discontinued or change price, and a recommendation block nobody has checked in a year is usually pointing at things you can no longer send.

Your shipping rules already influence order size whether you have thought about them or not. Every buyer does the arithmetic of whether the postage is worth it, and the rules you set decide whether that arithmetic pushes them up or sends them away.

The main tool is the free shipping threshold, which deserves a guide of its own because setting it badly is expensive in both directions. The short version is that the line belongs a little above where a typical order already lands, so it is reachable with one more item rather than three, and it needs to be set from your own order distribution rather than copied from anywhere else.

Beyond the threshold, three smaller rules do useful work. Flat rate shipping that does not increase with the number of items removes the reason to split an order into two, which is a real behaviour when postage is charged per item. A cheaper rate for click and collect or locker delivery lets you keep the margin on small orders without refusing them. And a shipping price that varies by weight or size, shown honestly on the product page, stops heavy low value items quietly eating the profit of everything else.

Where a threshold exists, the gap has to be visible at the moment it can change behaviour. On the product page beside the price, in the cart, and in the drawer when an item is added. Showing the gap in money is what works. A progress bar with no number is decoration.

One thing to avoid: setting the threshold above the point where most people can reasonably reach it. A line too far away is read as a message that shipping is expensive here, and it depresses orders instead of raising them. The threshold guide covers how to find the figure from your own data and how to test a change without guessing.

Some orders lose money and no amount of merchandising will fix them. Find them before you spend effort on the rest, because removing a loss is worth exactly as much as creating a gain and it is usually faster.

Work out your break even order value. Add the pick and pack time, the packaging, the outbound carrier cost, the payment fee and the average returns cost for that category, then find the order value at which contribution reaches zero. Now count how many of your orders fall below it. In stores with a large range of cheap accessories, that number surprises people.

You have four honest options. Set a minimum order value, which is blunt but clear. Add a small order fee, named plainly, which lets people choose. Sell the cheap items in packs so the smallest possible order is still viable. Or leave the loss in place deliberately, because those orders bring first time customers who buy properly later, which is a legitimate choice as long as it is a choice and you can see the repeat rate that justifies it.

Check the postage bands before you do anything else. Many stores lose money on small orders because a slightly oversized mailer pushes the parcel into a more expensive band. Changing the packaging can save more than any pricing rule, and it costs the customer nothing.

Also look at the free shipping threshold from this angle. If your threshold sits below your break even point, every order at exactly the threshold is a loss, and the more successful your threshold messaging is, the more of those you get.

When you do introduce a minimum or a fee, say it early. On the product page and the cart, not at the checkout. A fee revealed at the payment step reads as a trick and costs the order you were trying to protect.

Order value work is the easiest of the four levers to fake, so the measurement has to be stricter than the tactic. Four numbers together tell the truth: average order value, items per order, contribution per order, and the return rate for the orders affected.

Run them side by side for the same period. A bundle that raises order value and items per order while holding contribution steady has done nothing except move money around. A quantity discount that raises order value while the return rate climbs has bought you refunds and restocking. A threshold that raises order value while the number of orders falls may have raised the average by losing your small customers, which looks like progress in one column and a shrinking business in another.

Watch the mix effect too. If a campaign brings a different kind of traffic, the average will move without any customer behaving differently. This is why the split you built in the first chapter matters: compare like with like, new customers against new customers, one channel against the same channel.

Give each move a fair window. Order value responds faster than repeat purchase work but slower than a checkout fix, and a fortnight of orders on either side is a reasonable read for a busy store. Avoid judging during a sale, because a discount period distorts every number in this list at once.

Keep a dated log, one line per change, with the four numbers before and the four numbers after. Six months in, that log tells you which moves are worth extending to the rest of the catalogue and which ones quietly cost money. Without it, you will remember the tactics that were fun to build rather than the ones that paid.

Then pick the next move from the shortlist you made earlier, and run it the same way.