How to Measure Conversion Gains From Virtual Try-On

To measure conversion gains from virtual try-on, compare similar shoppers who can use it with shoppers who cannot, over the same dates, using one fixed definition of conversion. The cleanest method is an A/B test, where half of your product page visitors see the try-on widget and half do not.

This guide walks through the setup, the metrics worth tracking, and the traps that make results look better or worse than they are. It also covers tailors and made-to-measure businesses, who need a slightly different yardstick.

What should you measure before you switch on virtual try-on?

Record your baseline first. Without it, any later number has nothing to be compared against.

Pull at least a few weeks of data for the product pages where you plan to add try-on. Note these figures:

• Product page conversion rate (orders divided by sessions on that page)
• Add-to-cart rate
• Average order value, in one currency (GBP, NGN or USD, whichever you report in)
• Return rate, and the share of returns marked "didn't fit" or "wrong size"
• Size-related support messages, such as "does this run small?"

Pick your primary metric now. For most stores it is product page conversion rate. Choosing after you see the results invites you to pick whichever number looks best.

Why can't you just compare shoppers who use try-on with those who don't?

You can look at that comparison, but you should not trust it. Shoppers who choose to try something on are usually more interested in buying already. They have scrolled further, stayed longer and often arrived with intent.

This is called selection bias. If users convert at a higher rate than non-users, part of that gap would exist even without the widget. Treat the comparison as a useful clue, not as proof.

The fix is to decide who sees the widget before anyone interacts with it. That means random assignment.

How do you run a clean A/B test for virtual try-on?

Follow these steps:

1. Write a hypothesis. For example: "Showing virtual try-on on dress pages will raise product page conversion rate."
2. Choose the pages. Start with categories where fit is a common worry, such as fitted dresses, trousers or tailored jackets.
3. Split traffic randomly. Use an A/B testing tool, or your platform's experiment features, to show the widget to one group and hide it from the other. Keep the split stable, so the same visitor always sees the same version.
4. Change nothing else. Do not launch a sale, new photography or a price change on those pages mid-test.
5. Run full weeks. Shopping behaviour differs between weekdays and weekends, and between payday and mid-month. Run the test for at least two complete weekly cycles, longer if traffic is low.
6. Decide the stopping rule in advance. Use a sample size calculator before you start, and do not stop the moment the numbers look good.

If you cannot split traffic, there are weaker options. You can enable try-on on some products and keep similar products as a comparison group. You can also compare before and after dates. Both are easier to run but easier to distort, because season, promotions and ad spend change over time.

Which metrics show the real gains?

Conversion rate is the headline, but it is not the whole story. Virtual try-on is meant to reduce doubt about fit, so look for effects further down the funnel too.

Metric	What it tells you	What to watch for
Product page conversion rate	Whether more visitors buy	Your primary measure
Add-to-cart rate	Whether confidence rises earlier	A rise here without more orders may signal checkout issues
Average order value	Whether shoppers buy more or better items	Compare in a single currency
Return rate for fit reasons	Whether sizing is more accurate	Needs weeks or months to mature
Support enquiries about sizing	Whether doubt is falling	Tag messages consistently
Try-on engagement rate	How many visitors use the widget	Useful context, not a gain by itself

Returns lag behind orders. A test that ends after two weeks may not capture them, so check the return data again once the delivery and return windows have passed.

How long should you wait, and how do you avoid false wins?

Small stores often have the hardest time here. With few daily orders, a handful of extra sales can look like a big jump and then vanish the next week.

Keep these habits:

• Do not read results daily and react. Check on a schedule.
• Use a significance or confidence readout from your testing tool, and be honest when the result is inconclusive.
• Segment carefully. Mobile and desktop, new and returning shoppers, and UK and Nigerian visitors may behave differently. But the more segments you slice, the more likely one looks good by chance.
• Record anything unusual, such as a stock-out, a courier delay or a influencer post that drove odd traffic.

An inconclusive result is still useful. It may mean the widget needs better placement, or that fit is not your main barrier on that product.

How should tailors and made-to-measure businesses measure it?

If you sell bespoke or made-to-measure pieces, a product page conversion rate may not capture what matters. Your journey is usually enquiry, measurement, deposit, fitting and delivery.

Track these instead:

• Enquiry-to-order rate
• Share of orders that reach the deposit stage
• Time from first enquiry to deposit
• Alterations or remakes caused by measurement errors
• Share of clients who reorder

You need tidy records for any of this. ZeDaq.ai includes tools for tailors in every subscription: a client CRM with contact details, delivery address and fit preferences, orders with deposits, a delivery calendar with automatic reminders, and invoicing sent by email or WhatsApp. Measurements can be captured by auto-capture, voice-guided posing or manual entry, and clients do not need to install an app. If your history sits in old paper books, you can photograph a page and have it read into a digital profile. You can learn more on the tailors page.

How do you set up tracking on a Shopify store?

ZeDaq.ai is an API and embeddable widget that adds AI virtual try-on and body sizing to product pages, and it is available on the Shopify App Store. The widget speaks 10 languages and supports multiple currencies, including USD, NGN, GBP and EUR, which helps if you sell across markets.

For measurement, set up your events before launch:

• Mark product page views, add-to-cart and purchase events in your analytics tool.
• Add a custom event or tag for widget opens, if your setup allows it.
• Note the exact launch date and the pages included.
• Tag returns with a reason code at the point of return.

There is a free plan and a 3-day free trial with no card required, so you can check the widget on your theme and confirm your tracking works before you run a full test. Sign up here to try it.

Quick checklist before you start

• Baseline data collected for the chosen pages
• One primary metric chosen and written down
• Random split in place, or a clear comparison group
• No other major changes planned during the test
• Return reasons tagged consistently
• A fixed end date and a review date for returns

FAQs

How long should a virtual try-on test run?

Run at least two full weekly cycles, and longer if your store gets few orders. The right length depends on your traffic, so use a sample size calculator before you begin. Check return data again after your return window has closed.

What is a good conversion lift from virtual try-on?

There is no universal figure, and it varies by product type, audience and store. Judge your own result against your own baseline. A smaller lift that is statistically reliable is more useful than a big one from a few days of data.

Can I measure virtual try-on without an A/B testing tool?

Yes, but with less certainty. You can enable it on some products and compare them with similar products, or compare before and after dates. Keep notes on promotions, ad spend and seasonality, since they can distort the comparison.

Should I compare try-on users with non-users?

Use it only as supporting evidence. People who try an item on tend to be more interested in buying already, so the gap overstates the effect. Random assignment gives a fairer answer.

Does virtual try-on affect returns as well as conversion?

It may, because the aim is to help shoppers judge fit before buying. Measure it by tracking the share of returns with fit-related reasons. Give it enough time, as returns arrive after orders.