Production upload-kit study · August 5, 2026
Across 3 product-list videos, the generator recovered 20 of 22 published product markers.
We froze three unusually structured, product-rich YouTube videos, generated one upload kit for each, and compared every returned product, chapter, and tag with the creator-published markers.
The product result
Useful draft coverage—with two omissions to catch.
All 20 returned products mapped to a product in the published reference list. The generator still missed 2 of the 22 published product markers, so review remains part of the workflow.
- Reference set22
physical products named in the creator’s description chapters across three videos
- Returned & matched20
generated products matched the published list; none were unmatched
- Time-aligned18
were within ±20 seconds of the published section start
The misses
Two published products did not appear in product output.
Both still received generated chapter entries. Neither received a product link, which is exactly the kind of omission a creator should catch before copying a draft.
1Mii SafeFly Mini Bluetooth Transmitter
8 Travel Tech Essentials for Your Next Trip
Open marker on YouTube ↗Xiaomi Smart Band 9
7 Travel Tech Products for Minimalist Travel | Android Tech Gear
Open marker on YouTube ↗The rest of the kit
Chapters aligned closely; tags fit the limit.
Chapter comparison used the same creator-published section markers. Tag checks covered character count and case-insensitive duplication—not search demand, ranking, or relevance.
Video by video
One complete product list, two with one miss.
All source videos came from Pack Hacker’s travel-tech list format. Product alignment counts use the published section start, not an independently judged first spoken mention.
| Source video | Products | Times ±20s | Chapters | Tags |
|---|---|---|---|---|
| 8 Travel Tech Essentials for Your Next Trip ↗Pack Hacker · 2024-08-23 | 7/81 missed | 7/7products | 9/9aligned | 20397 characters |
| 7 Travel Tech Products for Minimalist Travel ↗Pack Hacker · 2025-03-07 | 7/7complete | 5/7products | 8/8aligned | 20436 characters |
| 7 Travel Tech Products for Minimalist Travel | Android Tech Gear ↗Pack Hacker · 2025-05-02 | 6/71 missed | 6/6products | 8/8aligned | 17368 characters |
Method
Frozen first, generated second.
The study is small by design, but its inputs, mappings, and calculations are inspectable. The committed research data contains every returned product, chapter, and tag from the three runs.
Free PDF report
Get the complete study.
Enter your email for immediate access to the findings, methodology, video-level results, and limitations.
We store your email to record report access and send an internal lead notification. Marketing messages are sent only if you select the optional checkbox. See our privacy notice.
- 01
Choose a favorable stress test
Three public, English-captioned videos from one creator, each with product names in published description chapters.
- 02
Freeze references before generation
Video IDs and all creator-published product and chapter markers were recorded before any output was inspected.
- 03
Run one normal preview per video
Each production request asked for products, chapters, and tags without a tracking ID. No video was swapped or rerun after seeing results.
- 04
Map identity, then calculate
Returned items were manually mapped by displayed product identity. A deterministic script validates mappings and recomputes the public summary.
Interpretation limits
What this study cannot tell us.
The result supports a narrow claim: on these three highly structured videos, the generator produced a strong draft that still required product and link review.
- All three videos come from one channel and one product-heavy format.
- Creator chapter markers are section starts, not manually adjudicated first-mention timestamps.
- The public response does not expose the deployed source revision or model identifier.
- Amazon listing identity, availability, price, commission eligibility, and conversion were not verified.
- The study does not measure editing time, search ranking, clicks, sales, or revenue.
Link safety is not link correctness.
The URL check confirmed only HTTPS, an Amazon host, and direct-product path syntax. It did not confirm that a page represented the exact spoken model, remained available, or was appropriate to publish.
Your video is the real test
Generate a draft, then keep the final say.
Start with one captioned YouTube video. Review every product, timestamp, chapter, and link before publishing.