Luxury virtual try-on and sizing evaluation scorecard

1. Fix the test conditions before the demo

Name a merchandising owner, a fit or technical-product owner, and an ecommerce owner. Have the brand’s privacy lead review photo handling. One person can hold more than one role, but record who makes each decision.

1. Define scope. Write down the collections, countries, devices and theme version included. Separate required capabilities from optional ones. Mark unsupported categories as out of scope rather than silently dropping them.
2. Build a representative sample. One practical starting point is 12 products across your main categories, including several difficult examples. This is a manageable screening exercise, not a statistically representative accuracy study. Expand the sample when failures cluster in a category.
3. Prepare product references. Save approved front and detail images, product and variant identifiers, colour, composition, intended fit and the relevant size chart. Record whether measurements describe the body or finished garment and which units apply.
4. Recruit consented participants. Use a range relevant to your customers, including different body proportions and photo conditions. Confirm permission for the test and storage of the evidence. Use anonymous test IDs in the worksheet; keep photos in an appropriately controlled location.
5. Set the decision rules. Identify essential garment details and unacceptable behaviour before seeing outputs. Agree who can approve a restricted launch and how a failed category will be excluded.

Prepare the size-chart checklist first if your product data is incomplete. Missing references should be recorded as an evidence gap.

2. Review the generated images against the garment

Use the same input set for each supplier. Record product ID, variant, participant ID, device, attempt number, completion time and evidence location. Retain first attempts and failed attempts as well as successful reruns.

• Product identity: check colour, print, pattern placement, buttons, trim, pockets and selected variant.
• Cut and construction: check neckline, lapels, seams, sleeve shape, hem and visible garment length. Note uncertainty where the reference photography is insufficient.
• Fabric appearance: check texture, transparency, sheen and ribbing. A still generated image cannot validate feel, stretch under movement or real drape.
• Shopper representation: inspect face, skin tone and proportions for unwanted changes.
• Outfits: where offered, confirm each piece remains identifiable and the pairing matches the selected catalogue items.
• Failure handling: record partial outputs, rejected photos, timeouts and retries. Report the denominator of attempted journeys alongside successful completions.

Have reviewers score the outputs independently before discussing disagreements. Where feasible, hide the supplier name during image review. Keep evidence for disagreements rather than averaging away a critical defect.

3. Test size recommendations independently

Agree a fitting reference with the team that understands the garment: an observed fitting with the physical item where available, or a documented assessment using the appropriate product measurements. Label desk-based assessments as such. A generated try-on image is not the fitting reference.

1. Record the relevant measurements, intended ease and participant fit preference in the controlled test record. The downloadable worksheet uses test IDs so it need not contain personal measurements.
2. Enter identical supported shopper inputs for each app and record the recommended product size, relevant explanation and any uncertainty displayed.
3. Compare the result with your agreed reference, including acceptable adjacent sizes if the fit preference allows them. Do not change the reference after seeing the app’s answer without documenting the reason.
4. Test boundary cases: between sizes, missing measurements, centimetres versus inches, an unavailable recommended variant, and a product with a different cut.
5. Record when the tool cannot make an appropriate recommendation and whether the shopper receives a useful fallback such as the size chart or human help.

Report agreement for the tested sample with the number of cases and reference method. A small screening test cannot justify a universal accuracy percentage across all customers or categories.

4. Rehearse the product-page journey

Test in a theme preview before a shopper-facing pilot. Run the full sequence on a phone and desktop: open the product, change variant, start the experience, upload an approved photo, receive the result, inspect size advice, close the panel and add the intended variant to the cart.

• Check brand styling, copy, whitespace and whether the feature obscures the size selector or purchase controls.
• Test keyboard navigation, visible focus, meaningful control labels, error messages and a screen-reader walkthrough.
• Record actual completion times and behaviour on a constrained connection; review waiting, retrying, cancellation and returning to shopping.
• Verify selected product and variant continuity after closing, reopening, reloading and navigating back.
• Confirm the intended behaviour for rejected inputs, service failure and plan limits. Record what was observed separately from what a supplier says will happen.
• Review the current privacy documentation and obtain answers about processing, retention, deletion, training use and contractual responsibilities. Record any unanswered requirement before launch approval.

For Rendered Fits, the Shopify listing describes photo try-on, size recommendations, outfit try-on and interface styling. This protocol verifies how the selected configuration performs in your store; it does not assume every requested control is available.

6. Separate launch readiness from business impact

Passing this protocol establishes readiness for the agreed pilot scope. It does not prove revenue uplift or fewer returns. Choose a primary commercial measure, define exposure consistently and compare equivalent groups where feasible.

Track eligible visitors, exposed visitors, starts, completed experiences, orders and mature returns. Report failure and abandonment rates as well as adoption. Include fees and operational effort when assessing contribution, and avoid treating influenced revenue as incremental revenue.

Use the measurement plan and worksheet for the next stage. Reopen quality testing when you change category, product imagery, sizing data, theme or app configuration.

Related buyer guidance

• Luxury fashion virtual try-on buyer’s guide — priorities and category-specific considerations.
• Compare Shopify virtual try-on apps — start a shortlist.
• Size recommendation guide — understand the separate sizing workflow.
• Rendered Fits product demonstration — see the shopper journey before testing your catalogue.

Method authored by the Rendered Fits team. Product capability reference: Rendered Fits Shopify App Store listing, checked 14 September 2026. The suggested sample size and scoring method are practical screening choices, not validated accuracy benchmarks.