Rendered Fits · Merchant guide
How to measure sizing and virtual try-on results
Separate feature use, purchases and mature returns. Use a clear measurement plan before claiming that sizing or try-on improved your store.
By the Rendered Fits team · Reviewed 13 September 2026
Start with the question you can answer
“Did shoppers use it?” and “Did it cause more profitable purchases?” need different evidence. An order after a try-on is an associated order. It is not automatically an extra order caused by the app. Define the observation window and the eligible products before evaluating results.
Use a consistent funnel
| Stage | Suggested definition | Do not confuse it with |
|---|---|---|
| Eligible visit | A visit to a product page included in the evaluation. | Every visit to the store. |
| Feature start | Shopper begins a sizing or try-on journey. | A completed recommendation or render. |
| Completed result | Recommendation shown or try-on successfully delivered. | A purchase or satisfied shopper. |
| Associated purchase | An order joined to eligible app activity within a stated window. | Incremental revenue. |
| Mature outcome | Kept, returned or exchanged item after a defined return window. | No return event yet. |
| Contribution | Revenue net of relevant product and operating costs. | Gross sales associated with app use. |
A useful comparison
Where feasible, randomly assign eligible traffic to comparable experiences and preserve the assignment. Evaluate all eligible visitors in each group, rather than only shoppers who chose to use the tool. Decide the duration, primary metric and decision rule in advance; avoid repeatedly stopping as soon as a positive result appears.
If a controlled test is not available, compare equivalent periods and document changes to promotions, traffic sources, stock, prices and seasonality. Treat the result as observational. Low traffic or few returns can make the result inconclusive even when the interface works well.
Worked example: percentage points matter
If 20 of 100 mature orders are returned in one group and 15 of 100 in another, the observed rates are 20% and 15%. The difference is 5 percentage points, or a 25% relative reduction. These are invented numbers to illustrate the calculation; they are not Rendered Fits results. The counts alone also do not establish statistical significance.
Record uncertainty and cost
Keep unjoined orders and pending returns visible. Specify whether the denominator is orders, items or shoppers and use it consistently. Include subscription fees, any processing charges, setup time and ongoing support. For a planning exercise, use the break-even calculator.
What we publish as evidence
We distinguish public merchant feedback, illustrative demonstrations and measured results. A service review is useful evidence about a merchant’s experience; it is not a controlled sales or returns study. Read the St Andrews merchant-evidence note.
See whether Rendered Fits suits your store
Try your product imagery, check size guidance and review the shopper journey before rolling out.