E-commerce Video ROI: Real Credit Costs, Break-Even Orders and Test Size
A product video pays off when the extra profit it brings in exceeds the full cost of making and running it. More views or higher revenue alone don't prove that. This worksheet shows how to price an AI product video down to the credit, work out how many extra orders you need to break even, and how much traffic a fair test actually needs.
Choose a look, add your product photo and review the credit cost.
Try a product video recipeEvidence note: The worked examples below use Viralance's real credit prices but invented sales figures. They are not a customer case study, a conversion-lift forecast or a guarantee.
What surveys say, and why it isn't your answer
In Wyzowl's State of Video Marketing 2026 survey, 82% of marketers said video gives them a good ROI, down from 93% the year before (Wyzowl, survey of 266 respondents in late 2025). That's useful context: most teams that use video think it's worth it. But it's self-reported, it covers all kinds of video, and it doesn't tell you whether your product page will sell more with a video on it. Only your own controlled comparison can answer that.
Step 1: Count the full cost of a usable asset
| Cost | What to record |
|---|---|
| Generation | Credits spent across all attempts, including rejected ones |
| Preparation | Photography, image cleanup and briefing time |
| Review and editing | Time spent checking labels, claims, sound and exports |
| Distribution | Extra ad spend or placement cost |
| Rework | Extra attempts after review |
Cost per usable asset = total production cost ÷ number of accepted assets.
If no asset passes review, report zero accepted assets and the money spent. Don't divide by zero or treat a rejected clip as free.
Step 2: Price the generation in credits and dollars
Credits come in one-time packs with no subscription and no expiry. The rate per credit depends on the pack you actually bought:
| Pack | Price | Credits | Per credit |
|---|---|---|---|
| Mini | $9 | 30 | $0.300 |
| Plus | $19 | 70 | $0.271 |
| Starter | $29 | 110 | $0.264 |
| Creator | $79 | 330 | $0.239 |
| Pro | $149 | 770 | $0.194 |
| Business | $299 | 1,650 | $0.181 |
A first purchase within 24 hours of signup adds 5–15% bonus credits depending on the pack, and later purchases add 10%. Bonus credits lower your effective rate; use the rate you really paid.
Typical product-video configurations (September 2026 catalog):
| Asset | Credits | On Mini | On Creator | On Business |
|---|---|---|---|---|
| Product Hero recipe (image restage + Hailuo 2.3 720p 6s) | 5 | $1.50 | $1.20 | $0.91 |
| Hailuo 2.3 image-to-video, 720p, 6s | 3 | $0.90 | $0.72 | $0.54 |
| Kling 3.0 Standard image-to-video, 5s, 1080p | 8 | $2.40 | $1.92 | $1.45 |
| VEO 3.1 Fast image-to-video, 720p, 8s | 12 | $3.60 | $2.87 | $2.17 |
| Seedance 2.0 UGC-style clip, 720p, 10s | 24 | $7.20 | $5.75 | $4.35 |
| Auto Subtitles on a finished clip | 3 | $0.90 | $0.72 | $0.54 |
Acceptance rate matters more than the sticker price. If you accept one Kling 3.0 Standard clip in three attempts, the generation cost of that asset is 24 credits, not 8. Record attempts from the start so you know your real rate per product type. Failed generations (errors, not clips you dislike) are refunded automatically, so don't count those.
Step 3: A worked example you can replace with your numbers
A shop makes videos for three products. For each it runs three Kling 3.0 Standard attempts (24 credits) and adds Auto Subtitles to the chosen clip (3 credits): 27 credits per product, 81 in total. Bought on the Creator pack, that's about $19.39 of credits.
Add two hours of photo prep and review at $25 an hour ($50) and $40 of extra ad spend to show the video. Total incremental cost: about $109.
Assume each extra order contributes $20 after product cost, fulfilment, transaction fees and expected returns.
- Break-even extra orders = $109 ÷ $20 = 5.5, so 6 orders.
- If a controlled test attributes 8 extra orders to the video, contribution is 8 × $20 = $160, net profit is $160 − $109 = $51, and ROI is $51 ÷ $109 ≈ 47%.
Notice what dominates: the credits are under a fifth of the total. Your time and ad spend usually matter more than the generation price, so don't let a cheap model tempt you into a long review cycle.
Step 4: Check you have enough traffic to measure anything
Small conversion changes need a lot of visitors to detect. Using the standard two-proportion test at 95% confidence and 80% power, the visitors you need in each group are roughly:
| Baseline conversion | With video | Visitors per group |
|---|---|---|
| 2.0% | 2.4% (a 20% relative lift) | ~21,100 |
| 2.0% | 2.5% (25% lift) | ~13,800 |
| 2.0% | 3.0% (50% lift) | ~3,800 |
| 3.0% | 3.6% (20% lift) | ~13,900 |
If a product page gets 1,000 visitors a week, a 20% lift on a 2% baseline would take about 42 weeks to detect reliably. For low-traffic stores it's usually better to test the video in paid ads, where you control the traffic, or to judge on the cost side only.
Step 5: Run a fair product-page test
- Choose a product with stable pricing and stock. Define the action you want to improve, such as completed purchases.
- Randomly split eligible visitors between the current page and the same page with the video. Don't change the offer, shipping promise or layout at the same time.
- Decide the duration and sample size before starting (Step 4), and don't stop early because the numbers look good.
- Compare conversion with its uncertainty, contribution and returns. Don't compare a promotion week with an ordinary week and credit the difference to video.
- Keep the original page live until the result supports a decision.
Add-to-cart rate and watch time help diagnose behaviour, but completed purchases and contribution decide the business result. Paid and organic visitors behave differently, so keep those segments separate.
Build a first product-video test
Open the Product Hero template (other product templates are in the template gallery), add your product photo and check the credit cost shown before generating. Start with one product so you can check packaging, labels and motion before spending on a batch. For ad-style formats such as testimonials and unboxings, see UGC ads.
Use the product-label checklist to reject misleading outputs and the three-shot plan to keep each asset focused. For model settings and costs for photo animation, see the image-to-video guide.
Copy this tracking sheet
Per asset: product; purpose; model and settings; attempts; credits deducted; pack rate paid; prep minutes; review minutes; accepted or rejected; reason; campaign or experiment ID.
Per experiment: dates; eligible visitors per group; orders per group; net sales; contribution per order; extra distribution cost; returns; decision and uncertainty. Keep customer-level details private when sharing findings.
When an AI product video isn't worth testing
- The product sells on exact detail (fit, texture, printed specs) that a generated video might misrepresent. Film it instead.
- The page gets too little traffic to measure a realistic lift (Step 4) and you have no paid channel to test in.
- Your contribution per order is near zero. No lift will cover the cost.
Frequently asked questions
What ROI should I expect from AI product videos?
There's no defensible universal figure. Surveys like Wyzowl's measure how marketers feel about video overall, not the lift on a given product page. Measure your own controlled comparison.
How much does one AI product video cost on Viralance?
From 3 credits ($0.54–$0.90 depending on pack) for a Hailuo 2.3 720p clip to 24 credits for a 10-second Seedance 2.0 clip. Multiply by the attempts you need per accepted asset.
Is revenue growth the same as ROI?
No. Revenue ignores costs and can rise for unrelated reasons. Use incremental contribution and include the full cost of the test.
Should every product get a video immediately?
Start with one product and a test you can measure, and expand when the workflow produces accurate assets and the results justify it.
