Find a buying question in the images you already have
PURO is a bamboo storage-box set from CALIDORO, our own brand. The PURO project presents the original and redesigned imagery alongside a reported Amazon Germany experiment. Here, we use its original set image to examine a creative decision.

Original imagery from the published PURO project, shown unchanged. Its German copy identifies the three-box set and explains handling and storage.
This image has several jobs: it introduces the three-box set, shows a person handling a box, and demonstrates nesting. The bottom-left panel contains all three boxes, while the other panels explain use. That gives the creative team a specific question: would a dedicated set-overview image make the contents easier to understand?
A useful alternative could put the complete set in one large composition and move the handling and nesting details to later images. Whether that improves purchasing behavior needs testing. It is a hypothesis drawn from the layout, not a result established by looking at the pictures.
Choose what you want to learn
Which image should appear first? | A suitable comparison | What it can tell you |
Does a set overview help? | A suitable comparison | What it can tell you |
Which visual direction should we adopt? | A suitable comparison | What it can tell you |
A broader test can be useful when the business decision is whether to adopt the whole redesign. A narrower test is more useful when you want to reuse a particular lesson. Name the scope before producing the alternatives.
A filled-in test brief for a three-box set
01 / Question What is included? Can shoppers recognize the complete set? | 02 / Alternative One clear view Test a dedicated overview in the same gallery slot. | 03 / Decision Use the test rule Assess purchases, uncertainty, and the practical change. |
The following example is for a store with a testing tool that supports replacing one gallery image. Check that capability before commissioning the variant.
Example test brief — proposed experiment
This is a proposed next test, not a reconstruction or result of the original PURO experiment. It assumes a testing tool that can replace one gallery image. Confirm that capability and test feasibility before production.
Customer question
Which boxes and lids are included in the set?
Hypothesis
A dedicated view of the complete set may make the offer easier to understand and improve purchases among visitors exposed to that gallery.
Control
The current multi-panel overview in its existing gallery position.
Alternative
One clear composition showing every included box and lid, with a concise contents label. Styling props must be distinguishable from included items.
Keep stable
The product and variant, price, offer, page copy, remaining gallery images, and the overview’s position.
Feasibility
Check the required sample and likely test duration with your testing provider, using your baseline conversion rate, eligible traffic, and the smallest improvement worth acting on.
Primary outcome
Purchase conversion rate, using the testing tool’s documented visitor and purchase definitions consistently for both groups.
Supporting checks
Revenue per visitor and add-to-cart rate, where measured reliably; note availability or promotional changes that could complicate the comparison.
Decision rule
Record the tool’s required evidence and stopping rule before launch. Adopt the alternative only when the results meet those criteria and the image passes product-accuracy review.
The same brief can be adapted to a cookware bundle, a bathroom accessory set, or another product with multiple included parts. Replace the buying question and confirmed contents; do not copy assumptions about PURO’s product or audience.
Approve both images before launching the test
A test cannot make an inaccurate image acceptable. Check every included part, label, material, and proportion against approved product references. Confirm that props do not imply an extra accessory and that any dimension or performance claim is supported.
Then inspect both variants at the size shoppers will actually see. The set should be understandable without zooming into small labels. Review the image in its intended gallery slot on desktop and mobile, including the thumbnail and any automatic crop. Apply the destination’s image requirements to each variant.
If the existing images leave those questions unresolved, request a free image review before expanding the production brief.
Use a controlled comparison and keep a record
Showing one image this month and another next month mixes the image change with changes in demand, traffic, and offers. Use a testing system that assigns comparable visitors to the alternatives during the same period and documents how that assignment works.
For example, Amazon’s Manage Your Experiments randomly divides customers into groups that see different content versions. Its available experiment types and eligibility determine what can be tested. A Shopify store needs a suitable experiment setup for its own page and traffic.
Save the exact image files and page settings used by each group, along with the hypothesis, dates, primary metric, sample information, and stopping rule. Follow the testing tool’s method; do not end a test simply because an early snapshot looks favorable.
Interpret the result at the level you tested
If a replacement-gallery test meets the planned decision rule, the result supports that gallery package in the tested context. You have not shown that a particular background caused the difference, that AI-assisted production is inherently better, or that another product will see the same result.
Read the estimated difference together with the tool’s uncertainty measure and the practical size of the change. Keep relative lift separate from percentage-point change. A result that does not meet the planned decision rule should remain inconclusive; it does not establish that the images are equally effective.
If traffic is too limited for a useful experiment, you can still check accuracy, readability, and whether shoppers understand the set. Treat that feedback as a usability check rather than a measured sales improvement.
Commission an image you can test
Bring the current image, the customer question, the proposed alternative, and the placement to your creative brief. That gives production a clear job and gives the test a result you can interpret.
To prepare new product-image variants, share your project details. Product in Frame develops the visuals; your testing setup measures how they perform with your customers.
