A bundle can solve a customer's purchase problem by combining everything needed for a use case. It can also increase order value while giving away margin on items the customer would have purchased at full price.
Design the test around the offer's mechanism and complete economics. The question is not simply whether bundle orders are larger. It is whether presenting this package creates a better business outcome than the defined alternative.
Use the average order value calculator to check basket revenue and the margin and markup calculator to inspect what remains after the included cost. A larger basket and a better contribution outcome are separate findings.
State why the bundle should help the customer
Choose a concrete hypothesis: reducing setup uncertainty, ensuring compatible accessories, making a gift easier to buy, or simplifying replenishment. Explain why the included products belong together.
Avoid starting with a discount percentage and searching for products to fill the box. A package that adds unwanted items may raise the price barrier, increase shipping costs, or produce more returns.
Describe the alternative the customer currently sees. The control could be the main product alone, separate add-ons, or an existing bundle. The test's interpretation depends on that comparison, so name it before production.
Define the offer and its catalog representation
Record each included item, quantity, variant rules, total price, discount conditions, and inventory dependency. Make the ad image and destination clear about what is included and what is shown only for illustration.
Google's bundle attribute documentation distinguishes merchant-created bundles containing a main product from manufacturer bundles and multipacks. Use the applicable specification for the actual package rather than labeling every multi-item offer identically.
Check product identifiers, images, titles, and destination behavior with the feed owner. A correctly described ad can still lead to a cart missing an accessory if the bundle implementation and catalog record disagree.
Calculate contribution before choosing the discount
Build the economics from the whole order. Include product costs, packaging, fulfillment, shipping subsidy, payment fees, and expected return costs under the business's model.
Consider this illustrative scenario:
| Offer | Net order revenue | Variable costs before ads | Contribution before ads |
|---|---|---|---|
| Main product | $80 | $42 | $38 |
| Bundle | $110 | $64 | $46 |
The bundle raises order revenue by $30 but contribution by $8 before advertising. If it increases acquisition cost by more than $8 per order, the contribution advantage disappears under these assumptions. The example does not establish the actual conversion response or incrementality.
Use the discount economics guide to examine how a larger discount changes the required sales response. Do not treat the advertised percentage saving as evidence that the offer is economically attractive.
Decide whether you are testing an offer or its presentation
If the bundle, price, creative, and destination all change, the test evaluates the complete package. That may be the right commercial question. It does not isolate whether the image, wording, or bundle composition caused the difference.
If the offer already exists and you want to test its presentation, keep the offer terms stable and vary the defined creative element. Use the offer-versus-visual test guide to keep these questions separate.
NIST's factorial-design overview explains designs that examine combinations of factors. Such designs require appropriate planning and data; they are not a reason to multiply ad variants beyond what the account can interpret.
Measure the basket, not only the advertised SKU
Track orders, bundle take rate, item quantities, net revenue, contribution, and relevant customer mix. Identify whether the bundle replaces separate purchases or brings genuinely additional items into the basket.
Keep a reliable mapping between the bundle's commercial identity and its component line items. A report that counts both a parent bundle and every component as separate revenue can overstate the result.
Separate new and returning customers where useful. Existing customers who already know the accessories may respond differently from first-time buyers who need more explanation. Avoid reading a customer-mix shift as pure offer improvement.
Watch fulfillment and returns during the test
One scarce component can constrain the entire bundle. Confirm how the store behaves when an included item is unavailable and whether substitutions require a revised customer promise.
Inspect support questions and return reasons. Customers may misunderstand quantities, compatibility, or the value of optional components. Those observations can explain why early order value looks strong while later contribution weakens.
Do not declare a bundle winner before the relevant outcome window matures. For a new package, early results may justify continued observation rather than a confident long-term return assumption.
Set a decision rule the team can use
Before launch, document the primary business metric, minimum useful improvement, maximum exposure, and conditions that would stop or invalidate the test. Base the plan on the account's baseline and uncertainty rather than a universal number of days or purchases.
Review both the result and the mechanism. If bundle adoption rises but overall contribution per visitor falls, inspect whether the higher price suppresses too many purchases. If contribution improves only among returning customers, the next test may focus on that audience or on clearer first-time education.
The final decision should state what was compared, what matured, what remained uncertain, and which version should run next. A successful bundle test produces a defensible offer decision, not merely a screenshot of a higher average order value.