TL;DR
- AR engagement varies wildly by category — eyewear/furniture/jewelry rings see real use; commodity products rarely do.
- Three categories drive meaningful AR: 'how does it look on me' (eyewear, jewelry), 'will it fit my space' (furniture), 'try-on' (footwear premium).
- Measure what matters: AR engagement rate, AR-to-purchase conversion vs photo-purchase, return rate on AR-purchased orders.
- AR is theatre when the answer it provides (color on a flat surface) isn't the customer's actual question.
- Decision check: if AR is on but no one uses it, you're paying the 1.7-1.9% per-item fee (per zakeke.zendesk.com) for a feature that isn't converting.
What AR preview actually does
Zakeke's AR preview lets the customer see the configured product in their physical space (or on their face for wearables) through their device camera before adding to cart. The customer points the camera; AR places the product at scale; the customer sees how it looks. For products where the unanswered question is 'how does this look in my context,' AR is a strong conversion tool. For products where the question is something else (price, materials, sizing accuracy), AR doesn't answer that question and engagement is low. See Zakeke 3D vs 2D for the broader framework on when 3D + AR earns its fee.
Where AR engagement is meaningful
- Eyewear: 'how do these frames look on my face' — AR's most-cited use case. Engagement is high; conversion lift is recognized. See Zakeke for Eyewear.
- Furniture: 'will this sofa fit my living room' — room-scale AR with the product at actual scale. High engagement on mid-to-high-AOV pieces. See Zakeke for Furniture.
- Jewelry rings: 'how does this ring look on my finger and at what size' — AR engagement is meaningful here too.
- Premium footwear: try-on AR for shoes is less mature than eyewear but engagement exists for premium configurations.
Where AR is theatre
- Commodity apparel: a customer doesn't need AR to imagine a basic tee on themselves — the product photo on a model does the job. AR engagement on commodity apparel is typically low.
- Personalized photo products: the customer's question is 'does my photo look right on the product,' which a flat preview answers. AR adds nothing.
- Small accessories: phone cases, keychains, stickers — AR scale is awkward and the questions AR answers don't apply.
- Color picks on small items: AR for color visualization on small items is theatre — color swatches on a mockup do the same job at no per-item fee.
What to measure to know if AR is working
| Metric | What it tells you | What 'working' looks like |
|---|---|---|
| AR engagement rate | % of product-page sessions that activate AR | Meaningful: 10%+ for AR-relevant categories. Below 2-3% suggests AR isn't the friction point. |
| AR-to-purchase conversion | Purchase rate of AR users vs non-AR users | AR users should convert higher; if equal, AR isn't doing work. |
| Return rate, AR vs non-AR purchases | Whether AR reduces 'not what I expected' returns | AR purchases should have lower 'didn't look right' returns. If similar, AR isn't reducing expectation gaps. |
| AR session duration | Whether customers actually explore the product in AR | 30+ second AR sessions suggest real interaction; sub-5-second suggests curiosity click without value. |
Run these metrics for at least 90 days post-AR-launch before deciding. AR conversion lift is sometimes immediate, sometimes builds as customers learn the feature exists; give it time before deactivating.
Decision: AR on or off?
- AR-relevant category + measurable engagement + conversion lift → keep AR, it earns the fee.
- AR-relevant category + low engagement → AR may not be discoverable. Surface it more prominently in the product page UX before deactivating.
- AR-relevant category + meaningful engagement + no conversion lift → investigate. AR is being used but not closing the sale. Possible UX issue or AR fidelity issue.
- Non-AR-relevant category (commodity apparel, photo products, small accessories) → AR is theatre. Consider whether the per-item fee is paying for features customers actually use.
AR not converting? Reassess the personalizer fit
If AR isn't doing real work for your category, you may be paying the per-item fee for a feature that isn't converting. For 2D personalization (text, photo, monogram) without AR overhead, a flat-fee personalizer fits — Print It My Way runs free, no per-item fees.
Install Print It My Way — Free Read when 3D justifies the fee →Frequently asked questions
Does Zakeke's AR preview actually lift conversion?
For AR-relevant categories — eyewear, furniture, jewelry rings, and to a lesser extent premium footwear — yes, AR can lift conversion meaningfully because it answers the dominant pre-purchase question ('how does this look on me' or 'will this fit my space'). For non-AR-relevant categories — commodity apparel, photo personalization products, small accessories — AR engagement is typically low and conversion lift is limited. The honest test is to measure AR engagement rate (% of sessions that activate AR) and AR-to-purchase conversion vs non-AR sessions on your specific catalog for 90 days, then decide. If AR isn't doing real work for your category, you're paying the per-item fee for a feature customers aren't using.
Which product categories show real AR engagement?
Three categories drive meaningful AR engagement: (1) 'how does it look on me' — eyewear (frame fit and face match), jewelry rings (size and look on finger), to a lesser extent premium footwear; (2) 'will it fit my space' — furniture (room-scale AR for sofas, sectionals, dining tables, beds); (3) premium try-on for products with high pre-purchase visualization uncertainty. Outside these categories AR is typically theatre — present, but engagement is low and conversion lift is minimal. The categories are the same ones Zakeke recommends in its own marketing because they're where AR genuinely solves the conversion problem.
What's AR engagement rate and what's meaningful?
AR engagement rate is the percentage of product-page sessions that activate AR — typically tracked by the personalizer's analytics or by event tracking integrated with GA4 or similar. Meaningful engagement for AR-relevant categories is 10%+ — meaning at least 1 in 10 product-page visitors activates AR. Below 2-3% suggests AR isn't discoverable (UX issue: customers don't know it's there) or isn't relevant to the customer's question. Above 20%+ on AR-relevant categories indicates AR is a primary part of the shopping flow. The benchmark depends on category; trial against your own baseline.
How do I tell if AR is reducing returns?
Compare return rates between AR-purchased and non-AR-purchased orders, segmented by return reason. AR purchases should show lower 'didn't look right' or 'didn't fit my space' returns than non-AR purchases. If the return rate is similar across both, AR isn't closing the expectation gap that drives those return reasons. The control is important — you're measuring whether AR specifically reduces visualization-related returns, not all returns. Run this analysis quarterly for AR-relevant categories so you have data to support keep-or-cut decisions on the AR feature.
Is AR worth the per-item fee?
Depends on your category and engagement metrics. Zakeke's per-item fee is in the 1.7-1.9% range per zakeke.zendesk.com on personalized orders, on top of plan subscriptions. The math: if AR genuinely converts users who would otherwise abandon (measurable as AR-to-purchase conversion lift), the fee pays back. If AR is on but engagement is low or AR users don't convert higher, the fee is paying for a feature that isn't earning. The honest decision is data-driven: measure on your store for at least 90 days, then decide. For non-AR-relevant categories, the answer is usually that AR isn't worth the fee and a 2D personalizer without per-item fees is the better fit.
What if AR engagement is low — is the AR broken or the category wrong?
Two possibilities, and the distinction matters. If AR is undiscoverable (the button is hidden, the UX doesn't surface AR's value), engagement will be low even on AR-relevant categories — the fix is product-page UX improvements (prominent AR button, AR demo in product description, AR-first product images). If the category isn't AR-relevant (commodity apparel, photo personalization), engagement will be low regardless of UX — the fix is to question whether AR makes sense for these products. Diagnose by testing prominent AR UX on the product page for 30 days; if engagement still doesn't move, the category isn't AR-relevant for your customers and the per-item fee is paying for theatre.