The most repeated advice about how to increase average order value is also the riskiest. Raise the free-shipping threshold, bundle three products together, drop an upsell into the cart, and watch the number climb.

The number does climb. What usually goes unchecked is whether the store made more money. A cart that grows from $60 to $78 because a shopper added a second size they'll send back, or because a sitewide code fired at checkout, can carry less contribution margin than the $60 order did. Shipping cost per parcel doesn't fall just because the subtotal rose.

So I judge every AOV tactic by a single measure: retained gross profit after returns, discounts, and checkout drop-off. Some levers pass that test easily. Others look strong in a dashboard and cost money. Here's which is which, and how to run the tests that tell them apart.

Table of Contents

• Which AOV Levers to Test First

  • Match the lever to your gap

• Protecting Margin When Order Values Rise

  • Sort the catalog before you push anything

  • Prefer bundle pricing to blanket codes

• How to Set a Free-Shipping Threshold That Grows the Cart

  • The five steps

• Building Bundles and Upsells Around Buying Intent

  • Establish the baseline first

  • Bundle around a use case, not a discount

  • Set the margin floor before you write the offer

• Cross-Sells and Checkout Add-Ons That Stay Out of the Way

  • Build the matrix from order lines

  • Placement decides whether it works

• Personalization and Quizzes That Match Each Shopper

  • Segment on signals you can actually observe

  • Guardrails before creative

  • Quizzes work when the questions are merchandising decisions

• Run Your First AOV Test This Month

Which AOV Levers to Test First

A bigger cart is a useful result only if the added revenue survives discounts, returns, and checkout friction. Compare each lever by its likely effect, where it fits in the journey, and what could erode the gain.

Treat the figures as rough comparisons, not forecasts. They come from different benchmark compilations and do not add together: a threshold, bundle, and upsell compete for the same opportunities to increase the cart. The recommendation figure measures revenue, not AOV, so it is not directly comparable. For discounts, compare the test group with a control and count the result only if it clears your margin floor.

Match the lever to your gap

The biggest reported effect is not necessarily the best first test. If orders already contain two or three units, bundles may have limited room to grow. If shoppers leave at the shipping step, changing an upsell placement will not address the problem.

Use your own order mix and funnel to narrow the choice:

1. What's your current AOV, and how wide is the spread? Compare the median and 75th percentile with the average. A few large orders can conceal a large group of smaller ones.

2. What share of orders already include a second item? A high share leaves less room for bundles and cross-sells, so consider another lever.

3. Where do shoppers leave? A drop at the cart, shipping step, or payment stage points to different friction. A shipping-step drop suggests investigating the threshold or how shipping costs appear before checkout.

The strongest first test addresses a clear gap and can be judged against a control, not just a higher cart subtotal.

Protecting Margin When Order Values Rise

A higher AOV can leave the business worse off, and the gap shows up weeks later when the parcel comes back. Ecommerce returns average roughly 16.9% of sales, apparel runs higher, and return processing can consume 20 to 30% of the original order value, according to Digital Applied's ecommerce statistics compilation. A cart padded with low-intent add-ons, poor product fit, or a promotion that encouraged over-ordering cuts contribution margin while the AOV chart still climbs.

Track five numbers beside AOV from day one: contribution margin per order, return-adjusted revenue, discount rate, shipping cost per order, and customer lifetime value. A lever that improves AOV and worsens any two of those needs a second look before it ships.

The guardrails worth keeping in front of you are all on one page.

Sort the catalog before you push anything

Consumables, replacement parts and accessories carry low return risk and absorb almost any bundling or cross-sell you point at them. Fit-sensitive goods behave the opposite way, whether that's apparel sizing, shade-matched cosmetics, or anything where the shopper is guessing. Suppress those in recommendations until you have return data at product level, and let the low-risk items do the cart-growing work.

Prefer bundle pricing to blanket codes

A threshold tied to a sitewide percentage code trains customers to wait for the code and cuts margin on every order it touches. Bundle pricing gives something real in return for the extra spend, and the discount stays contained inside the bundle. Quantity breaks such as buy two and save 10% belong only on consumables and refill sets where return risk is low, and each break has to clear your margin floor before it goes live.

The question that settles most AOV arguments inside a brand: does this tactic increase retained gross profit after returns and fulfillment, or only the checkout subtotal?

One sitewide threshold rarely fits every category, so split results by category, acquisition source, and return behavior. International stores have an extra reason to split: duties, regional shipping rates, and currency differences make a single global number misleading. Getting the underlying retail pricing strategies right is what makes a margin floor survivable under discounting.

How to Set a Free-Shipping Threshold That Grows the Cart

The threshold is the cheapest lever to test and the easiest to set badly. Get the number right and shoppers add a product to avoid a delivery fee. Get it wrong and they leave at the cart page without telling you why.

Here's the sequence I'd run in any store, regardless of catalog.

Start with market context as a sanity check. A 2025 summary of ecommerce research from Carriyo reports that 93% of consumers take some action to qualify for free shipping, with purchasing additional items the most common behavior, and associates threshold-based free shipping with a 15 to 20% increase in AOV. The same compilation puts the typical US retailer's threshold at $64 in 2023, up 23.1% from $52 in 2019. None of that is your number. Your cart data is.

If you want to see the mechanic working before you change a setting, this walkthrough covers the cart-level version.

The five steps

1. Pull 60 to 90 days of orders and find the median cart value. The average gets dragged upward by a handful of large orders, and a threshold built on the average lands above what most shoppers spend.

2. Set the starting threshold 20 to 25% above current AOV. A store averaging $60 would test $72 to $75. That's close enough to reach in one addition and far enough to require a real one.

3. Show the exact remaining amount early, in the cart and on a progress bar. Shoppers who discover the gap at the shipping step feel tricked. Shoppers who see it at add-to-cart treat it as a target.

4. Recommend one add-on that closes the gap. A low-cost refill, a consumable, or an accessory. A generic popular-items carousel gives them a reason to browse, not a reason to add.

5. Run the threshold against a control and read four metrics. Conversion rate, gross margin, shipping cost, and revenue per visitor. AOV alone will always approve a threshold that's too high.

The failure signature is easy to spot once you know it. AOV rises, conversion falls, and revenue per visitor goes flat or down. That's sticker shock, and the fix is to lower the number, not to add louder messaging.

Keep the language honest too. If exclusions apply or delivery slows above the threshold, say so in the same sentence, because a customer who hits the number and still pays for shipping won't try twice.

Building Bundles and Upsells Around Buying Intent

Bundles lift AOV by roughly 20 to 35% on average and upsells by 10 to 30%, depending on placement and relevance, according to Ringly's 2026 ecommerce benchmark compilation. The same compilation cites a best case of a 55% AOV increase from deliberate bundling. Treat that as the ceiling for well-executed work, not the result you should budget for.

Before any of it goes live, you need a number to compare against.

Establish the baseline first

Record AOV, units per transaction, bundle attach rate, conversion rate, and margin for several weeks before you change a page. Without that window, a bundle that lifts AOV by 20% and cuts conversion by 8% looks like a win in a monthly report. A $50 single-product order that becomes a $60 bundle is a good outcome only if conversion and contribution margin hold.

Bundle around a use case, not a discount

The bundles that survive are the ones a customer would have assembled anyway: a starter kit for someone new to the category, a routine built around a regimen, a refill set for a repeat buyer, an accessory the main product needs in order to work. Bundling two unrelated items and cutting the price is a discount wearing a costume, and it teaches the customer to expect the costume next time.

A bundle that needs a discount to make sense is a discount, and the margin comes back out of next quarter.

Set the margin floor before you write the offer

Decide the minimum contribution margin a bundle can carry, then work backwards to the price. If the discount pushes the bundle below the floor, cut an item rather than cut the price again. Test against a control that shows the same products separately, or against a post-add-to-cart recommendation, so you can tell whether the bundle did the work or the traffic did.

Three failure modes repeat across stores: forced bundles the shopper can't decline, offers that repeat what they bought last month, and upsells for a product already in the cart. More structures that hold up under testing are collected in these product bundling examples.

Cross-Sells and Checkout Add-Ons That Stay Out of the Way

An add-on can raise the cart subtotal and still reduce retained gross profit. If it slows checkout or adds a discounted item that gets returned, the bigger order may be the less profitable one.

Build the matrix from order lines

Start with order-line data. Support is the share of orders containing an item. Confidence is the probability that a shopper buys item B given that they bought item A. Lift is confidence divided by B's baseline purchase rate. A lift of 2 means the pair sells together twice as often as chance would suggest.

Use these measures to find pairings worth testing, then check whether their margin holds up after discounts and returns. High sales volume alone does not prove that two products belong together.

Placement decides whether it works

Match placement to the shopper's decision. Product pages can explain a related item; a cart drawer can show a simple complement; a post-purchase offer can present an add-on without putting the current order at risk. Baymard's checkout usability research, based on a benchmark of 344 leading US and European ecommerce sites and more than 30,000 manually reviewed checkout elements, warns that requiring shoppers to accept or reject an offer before proceeding can feel aggressive. In one test, 66% of users forced through a separate Amazon cross-sell step showed extreme frustration.

Cart abandonment is common enough that even minor checkout friction can cost more revenue than an add-on gains. Keep recommendations passive and dismissible, and make the pay button the clearest action on the screen. Check that an offer does not conflict with the free-shipping threshold or promote unavailable inventory.

For Acure, Presidio refined mobile navigation and cart flows to increase merchandising visibility and reduce drop-off. The same principle applies to individual offers: make them easy to notice, but never a hurdle to completing the order.

Judge the test by more than attach rate:

• Attach rate shows whether shoppers take the pairing.

• Incremental AOV shows whether it raises order value against a holdout.

• Checkout completion catches friction the added revenue can conceal.

• Discount-adjusted contribution margin accounts for the cost of the offer.

• Net revenue per initiated checkout includes sessions that did not become orders.

Personalization and Quizzes That Match Each Shopper

Personalization can lift cart value while lowering profit if recommendations push discounted or frequently returned products. Judge each program by retained gross profit after returns, discounts, and checkout drop-off, not by the subtotal alone. A popular-products widget is easy to add; it still needs to change what shoppers buy to earn its space.

Segment on signals you can actually observe

Use observable behavior: products viewed, cart contents, purchase history, traffic source, geography, and lifecycle stage. Give each segment one recommendation objective, then place it where the shopper can act on it.

McKinsey's analysis of personalization value reports a 10 to 15% revenue lift most often, with results ranging from 5 to 25% depending on sector and execution quality. It also reports that faster-growing companies derive 40% more revenue from personalization than slower-growing peers. These are revenue figures, not AOV forecasts; the wide range makes the case for testing rather than assuming a lift.

Guardrails before creative

Set recommendation rules before designing the widget. Check inventory, enforce a margin floor and price ceiling, and suppress products the customer already owns. Then compare randomized traffic against a non-personalized control. Track add-to-cart rate, conversion rate, AOV, gross margin per session, and revenue per visitor, split by new versus returning shoppers and device.

Quizzes work when the questions are merchandising decisions

A quiz should ask questions that lead to useful product choices. Ask about the shopper's use case, constraints, and experience level, then map answers to relevant recommendations. Fellow reported that its quiz boosted AOV by 28% in a week. Treat that as one brand's result, not a benchmark or a result attributable to another product. Presidio's Quiz Kit is an AI-powered quiz app designed to help boost average order value with smarter recommendations.

Two pitfalls can erase the gain. Optimizing click-through rate may favor an attention-grabbing recommendation that lowers checkout completion or margin. Personalizing from thin data can make offers feel random or repetitive. Judge the program by incremental revenue per visitor and retained gross profit, including the effects of returns and discounts; keep the control group running long enough to account for delayed purchases and repeat orders.

Run Your First AOV Test This Month

Set the calendar before you set the variant. Five steps, no new tooling required.

1. Baseline. For four weeks starting, say, 16 November, record AOV, units per order, conversion rate, and gross margin. Change nothing during that window.

2. Pick one lever from the first table that matches your largest gap. A low bundle attach rate points to bundles, cart-page drop-off points to thresholds, and weak product pages point to placement.

3. Build a control and a variant, and hold out a share of traffic so the comparison stays clean and repeat buyers don't contaminate both sides.

4. Run for several weeks, long enough to collect a readable number of orders, then check new and returning shoppers separately.

5. Judge on net revenue per initiated checkout and contribution margin after returns and discounts. Decide before the test starts which metric ends it.

The kill rule comes first. If the variant wins on AOV and loses on contribution margin, roll it back and test a lower-risk placement instead of defending the number. This is the discipline behind our Shopify A/B testing work: one variable, a clean holdout, a decision rule agreed in advance. We ran a structured conversion optimization program with Acure around mobile performance, PDP clarity, and long-term DTC growth, and the order of work was always the same. Fix the path, then add the offer.

AOV gains hold up when the storefront, the offer logic, and the measurement are built together rather than bolted on one at a time. Presidio builds and optimizes Shopify storefronts, apps, and quizzes for DTC brands, and runs the CRO, personalization, and testing programs that keep an AOV experiment honest. If you want your threshold, bundle structure, and checkout flow reviewed by people who do this every week, start at Presidio.