Photo vs Video vs Text Reviews: What the Conversion Data Actually Shows

The "which review format converts best" question gets recycled every year, with each review app vendor publishing a study claiming their format is the winner. This post pulls together the actual conversion data (from public studies, our own aggregate data across stores using Eevy AI's genetic algorithm to test review formats, and the experimental literature on visual social proof) to give you a defensible answer for your store.

The short version: video reviews convert highest per impression, but photo reviews have the best ratio of conversion lift to collection difficulty, and text reviews remain the most-volume-efficient format. The right mix depends on your category, AOV, and current review velocity.

What "Converts Best" Actually Means

Before quoting numbers, the question itself needs unpacking. "Conversion lift from review format" can mean three different things:

1. Display-level lift: The difference in CVR between a product page showing photo/video reviews vs one showing only text reviews of the same products
2. Per-review lift: The marginal CVR contribution of adding one more photo review vs one more text review
3. Collection-cost-adjusted lift: Display lift divided by the difficulty of collecting that format

Most vendor studies report (1), the gross display-level number, without adjusting for the fact that getting customers to submit a photo review is 5-10x harder than getting them to submit a text review, and getting a video review is 5-10x harder still. When you factor in collection cost, the rank order can flip.

The Aggregate Display-Level Numbers

Across the public studies (Yotpo's 2024 review report, Loox's annual benchmark, Reviews.io merchant survey, Bazaarvoice consumer behavior study) and our own aggregate test data from stores running review-format A/B tests in Eevy AI, the consistent display-level findings are:

• Pages showing video reviews convert 2.4-4.1x higher than pages showing only text reviews on the same products
• Pages showing photo reviews convert 1.6-2.3x higher than pages showing only text reviews on the same products
• Mixed displays (photo + text + video) convert 2.0-3.0x higher than text-only displays
• The gap widens with AOV. On products under $30, the lift from photo or video is 1.3-1.8x. On products over $200, the lift can exceed 4x.

These numbers vary by category. Apparel and beauty see the largest video lift because shoppers want to see real people wearing or using the product. Consumables (food, supplements, cleaning products) see the smallest lift because the visual differentiation is low. Technical products (electronics, tools) see large photo lift but moderate video lift.

Why Video Wins Per Impression

The mechanism is psychological rather than informational. Research on social proof and product perception (notably Bazaarvoice's 2023 consumer survey of 9,000 shoppers) finds:

• Authenticity inference: Video reviews are perceived as harder to fake than text reviews. Shoppers consciously and unconsciously discount text reviews because they know they can be written by anyone. Video carries a credibility premium.
• Use-case visualization: A 30-second video shows a product in use in a way no text or photo can. Shoppers learn fit, scale, behavior, and edge cases.
• Time investment signal: A reviewer who took the time to record video signals genuine satisfaction. This is a strong status signal for the product.

The same survey found that 79% of shoppers say they trust video reviews more than text-only, and 64% say they want to see at least one video before buying products over $100.

Why Photo Reviews Win on Cost-Adjusted Lift

Photo reviews are the underrated middle path. Their display lift is 60-80% as large as video's, but the collection cost is roughly 5-10x lower:

• Submission barrier: A shopper takes 10 seconds to attach a phone photo. A video requires lighting, framing, and 1-3 minutes of effort.
• Review-app prompts work: Standard post-purchase email/SMS flows that ask for photos get response rates of 15-30% on average, vs 2-7% for video.
• Editing-free: Most shoppers will not edit a video before submitting, but un-edited video looks lower-quality. Photos are good as-is.

The result is that a store can typically build a robust photo-review corpus 3-5x faster than a video-review corpus, and the extra volume of photo reviews offsets video's per-review edge.

If you are starting from zero and building review velocity, prioritize photo reviews first. Layer in video review collection 6-12 months later when the operational cost of video review prompts has fallen.

When Text Reviews Still Matter Most

Text-only reviews are not dead. They serve three roles that photo and video cannot:

1. AI Overview and rich snippet input: Google's AI Overview feature, the rich-snippet star ratings on SERPs, and most schema-based review surfaces all extract from text. Photo and video reviews enrich the on-site display but do not directly affect SERP visibility.
2. Specific question answering: A shopper asking "is this good for tall people" needs a review that says yes or no in words. Photos and videos rarely answer specific questions explicitly.
3. Volume signal: The displayed review count (e.g., "Based on 1,247 reviews") drives more conversion than the format of the underlying reviews. High volume of text reviews is better for conversion than low volume of perfect video reviews.

Most stores should aim for a corpus that is roughly 65-75% text, 20-30% photo, 5-10% video. The exact mix depends on category and AOV.

The Review Display Format Decision

The conversion data answers a separate question from the review collection decision. Even if your underlying corpus is 75% text, you still need to decide what gets displayed prominently on the product page.

The empirical answer from genetic-algorithm A/B testing on Eevy AI is consistent: lead with photo and video reviews, but always show enough text to support specific shopper questions. Specifically:

• Above the fold or in the buy-box area, prioritize photo and video tiles
• Show 3-5 most-recent or highest-engagement photo/video reviews as a carousel or grid
• Below the fold, show a paginated text-review list with filtering by rating, length, and verified-purchase status
• Reserve a "highlighted reviews" slot for AI-summarized review content (synthesized pros and cons across the corpus)

Genetic-algorithm testing finds that the lift from this layout vs a chronological text-only list ranges from 8% to 22% on revenue per visitor, depending on category.

What This Means for Your Store

If you are reading this and trying to decide where to put effort, the practical priority list is:

1. Volume first: Get to 100+ reviews on your top 20 products before optimizing format mix. Below 100 reviews, the social proof signal is too thin to justify worrying about format.
2. Photo collection second: Set up a post-purchase flow that asks for a photo with the review. Aim for 20-30% photo attachment rate on submitted reviews.
3. Display optimization third: Test photo-led layouts vs text-led layouts on your product pages. Use a self-optimizing review section if available, otherwise A/B test manually with significant sample size.
4. Video collection last: Once your photo flow is working, layer in video review prompts on your top 10 products. Even modest video corpus (5-15 videos per product) drives meaningful lift.

The mistake most stores make is jumping to video first. It is the highest-converting format per impression, but the collection investment dominates the return for the first 6-12 months.

How Eevy AI Approaches the Format Mix

Eevy AI does not pick a format winner up front. Our genetic algorithm tests photo-led, video-led, text-led, and mixed display layouts against each other on your real traffic, evolves the layout that converts best for your specific category and AOV, and continues testing as your corpus grows.

For a store with 50 reviews and 5 photos, the optimal layout is different from the same store at 500 reviews and 80 photos. The right layout also changes as visitor traffic changes; desktop visitors and mobile visitors respond to different formats, and even time-of-day matters in some categories.

The result is that you do not have to guess at the format mix. You collect what you can collect, you let the algorithm decide what to display when, and you keep your time on actually growing review volume rather than running A/B tests by hand.