What the Virtual Try-On Statistics Actually Say
The widely quoted 94% conversion lift is not what it looks like. Here is where the virtual try-on numbers come from, which ones survive scrutiny, and which are recycled.
If you have spent an afternoon researching whether virtual try-on is worth building, you have met the same wall everyone meets: a hundred pages quoting a handful of percentages, none of which link to anything you can check.
We went and traced them. The short version is that the most-quoted number in the category does not mean what it is used to mean, several others appear to have been invented, and the genuinely solid research measures something different from what buyers assume.
Where the 94% actually comes from
Start with the statistic you have seen most: virtual try-on increases conversion by 94%.
Follow the citations and they converge on one place. Snap's business blog attributes it to a 2020 Harvard Business Review article. HBR was reporting figures from Shopify. Shopify's own phrasing, in their post on 3D commerce and a 2020 post on X, is this:
Interactions with products having 3D/AR content showed a 94% higher conversion rate than for products without AR/3D.
Read that carefully, because the wording is doing a lot of work.
It is platform-wide merchant data. No sample size, no methodology, no control group, no publication. And critically, it describes interactions — it compares shoppers who engaged with AR content against shoppers who did not.
The self-selection problem
This is the flaw underneath nearly every conversion statistic in the category, and it is worth understanding properly, because it will be the first thing a skeptical CFO asks.
A shopper who opens a try-on tool is not a random shopper. They have already found the product, stayed on the page, and taken a deliberate action to see the item on themselves. That person was closer to buying before they touched the AR button.
So when you compare them against all other visitors — a pool that includes accidental clicks, price-checkers, and people who bounced in four seconds — you are measuring intent at least as much as you are measuring the tool.
The honest position is that AR try-on plausibly does lift conversion, and the effect could be substantial — but the published evidence does not establish it, and quoting 94% as though it did is a claim the source itself does not make.
What holds up
Strip out the marketing and a real body of evidence remains. It is mostly about confidence rather than transactions, which is less exciting and considerably more defensible.
The strongest widely available source is a 2021 study by Deloitte Digital commissioned by Snap Inc. Its methodology is published: 15,000 consumers across 15 countries, a 20-minute online survey, plus more than 50 interviews with AR practitioners.
15,000
Consumers surveyed, 15 countries
56%
Say AR builds confidence in product quality
41%
More likely to consider a brand offering AR
51%
Willing to use AR to assess products
The first three come from the Deloitte Digital and Snap study; the last from a 2019 Nielsen global survey, which found AR and VR topped the list of technologies consumers wanted help from in daily life.
Add to that the case-study evidence, which is weaker but at least attributable. Shopify published results from Rebecca Minkoff, who put 3D models and AR on around fifty product pages: shoppers viewing a product as a 3D model were 44% more likely to add it to cart, and those using AR were 65% more likely to purchase.
That is one brand, one category, no control group. Treat it as a signal, not a benchmark — but it is a real, named, checkable result, which puts it well ahead of most of what circulates.
| Source | What it measures | How much weight to give it |
|---|---|---|
| Deloitte Digital / Snap (2021) | Stated confidence and brand consideration | High — large sample, published method |
| Nielsen (2019) | Willingness to use AR to assess products | High — independent, large sample |
| Shopify / Rebecca Minkoff | Add-to-cart and purchase, one brand | Medium — attributable, no control |
| Shopify "94%" | Conversion among AR-interacting sessions | Low — no method, self-selected |
| Vendor blog figures | Varies | None — unattributed marketing |
What to throw out
Some of what is circulating is not weak evidence. It is not evidence.
The "2024 McKinsey study" showing an 18% lift. Several content-marketing sites cite this. We could not locate any such study. It has the texture of a figure that was invented once and then copied, and it now appears confidently in dozens of posts.
"2.5x conversion" and "up to 250%". These trace only to vendor product pages. No methodology, no sample, no independent publication.
Trade headlines with nothing underneath. A 2025 industry article titled around eyewear retailers using try-on to boost conversion rates contains, on inspection, no numbers at all — only that customers using try-on are "considerably more likely to purchase."
Academic work does exist, but it is not the rescue it looks like. A 2024 PLOS ONE paper on AR shopping and purchase intention surveyed 202 people and measured stated intent, not sales; its model fit statistics fall short of conventional thresholds. Interesting, not citable as proof.
How to get a number you can trust
The good news is that the measurement you need is not hard, and it beats every published statistic because it is about your products and your traffic.
Split at the page, not the user
Randomly assign product-page traffic into two groups. One sees the try-on entry point, the other does not. Do not filter by who clicked — that recreates the exact bias this article is about.
Measure the whole group
Compare conversion across all visitors in each arm, including the ones who ignored the widget. That gives you the effect of offering try-on, which is the only decision you actually control.
Run it long enough to mean something
Ecommerce conversion is noisy. A few days of data on a low-traffic page will produce a confident-looking result that reverses next week. Decide your sample size before you start.
Watch the secondary signals
Time on product page, add-to-cart rate, and return rate often move earlier and more visibly than conversion, and they tell you whether shoppers are getting genuine information from it.
If you sell eyewear or jewelry, the mechanism you are testing is specific and worth naming. Shoppers hesitate because a photo cannot answer a question they have about themselves — will these frames suit my face, how big is this stone really on my hand. We have written about both of those failures in detail: why a ring never looks the same on your hand as in the photo and the three questions every glasses shopper has.
What we would say to a retailer
Virtual try-on has a real and well-documented effect on how confident shoppers feel. That is the finding the evidence supports, and it is not a small one — confidence is what stands between a full cart and an abandoned one, particularly in categories where the item has to suit the person.
What the evidence does not support is a specific conversion multiplier, from us or from anyone else. Any vendor quoting you one, including a lift they attribute to their own platform, should be asked the three questions above.
The practical path is to install it, put it where the hesitation actually happens — the product page, at the moment of comparing two similar items — and split-test it against your own baseline. The documentation covers embedding it on a product page, and the platform ships try-on, IPD measurement, face shape detection, and the 3D viewer together, so a single test can cover more than one hypothesis.
Then you will have a statistic worth quoting. It will be yours, it will be about your products, and you will be able to answer all three questions about it.
Frequently asked questions
Does virtual try-on increase conversion rate?
There is good evidence that it increases purchase confidence, and case-study evidence that it correlates with higher conversion. There is no published independent study establishing a causal conversion lift with a control group. Most quoted figures compare shoppers who chose to use try-on against everyone else, which measures intent as much as it measures the tool.
Where does the 94% conversion statistic come from?
It originates with Shopify, describing merchants whose products carry 3D or AR content. The precise wording is that interactions with those products showed a 94% higher conversion rate. It is internal platform data, published without methodology or sample size, and it compares people who interacted with AR against people who did not.
Why are virtual try-on statistics unreliable?
Three reasons. Most are vendor marketing with no methodology attached. Many compare self-selected users against all visitors, which confuses engagement with causation. And a large number are simply copied between blog posts until the original source is untraceable or was never real.
What AR retail research is actually credible?
The 2021 Deloitte Digital study commissioned by Snap Inc. is the strongest widely available source, surveying 15,000 consumers across 15 countries with published methodology. Nielsen has run comparable consumer surveys. Both measure attitudes and stated intent rather than transactions, which is a real limitation but an honestly stated one.
How can a retailer measure virtual try-on impact properly?
Run a split test at the product-page level rather than comparing users to non-users. Show the try-on entry point to a random half of traffic and hide it from the other half, then compare conversion across the whole group. That measures the effect of offering try-on, which is the decision you actually control.
