Guides
Prepare an Ecommerce Catalog for Repeatable Product Videos
Build a reliable SKU ledger, map products to accurate scenes, preview representative items, and check every output against current catalog facts.

In this article
01 / PRODUCT DATA
Build a SKU ledger before creating a video
A catalog video workflow begins with reliable product records, not with a prompt. Use one row per sellable variant and identify the source of every fact. A missing capacity, old offer date, or mismatched image can multiply into many incorrect deliverables if the batch runs before review.
For a clearly fictional example, a small shop has three ceramic travel cup variants: blue, sand, and charcoal. Its ledger records each SKU, color, 350 ml capacity, approved product photo, current price, offer end date, market, and destination URL. Those are example values for this workflow, not claims about a real item.
- Give each variant a stable SKU and a current product image.
- Separate confirmed product facts from promotional language.
- Mark prices, discounts, inventory, and offer dates with an owner and last-checked time.
- Store approved logos, typography, and disclaimers alongside the campaign brief.
02 / TEMPLATE DESIGN
Map each product to a scene that can show it accurately
Choose a scene only after checking how the product will appear in it. A plain product turntable may work for similar cup variants; a lifestyle scene with a hand, lid, or pouring action may need a separate reference and closer review. Keep product shape and scale visible, and avoid relying on generated lettering for exact prices or claims.
- Group items by shape and use, not only by the catalog category name.
- Pick one approved visual structure for each group.
- Define which fields can change automatically: image, item name, color, offer, or call to action.
- Keep legal copy and exact offer text as controlled overlays where practical.
- Preview one representative SKU from each group before expanding the batch.
03 / PILOT RUN
Run a small batch and verify every SKU
Start with a pilot of the three fictional cup variants. If each needs a square and vertical deliverable, that is 3 SKUs × 2 formats = 6 outputs. The count describes the review workload; it does not guarantee that any tool can generate them in one batch. Record which source data, scene, and editor produced each file.
- Compare product silhouette, color, markings, packaging, and visible accessories with the approved source.
- Check every price, discount, date, disclaimer, and link against the live product record.
- Confirm the crop shows the relevant product detail in each intended format.
- Have a second reviewer approve any claim, human likeness, voice, or market-specific copy.
The source article describes 11Labs AI-specific template, API, and editing workflows. Treat those as publisher claims and confirm current official documentation before relying on a provider’s import, status tracking, or batch features.
04 / OPERATIONS
Track failures and release only approved files
Use an output ledger with SKU, source-record version, format, job identifier if available, attempt number, reviewer, status, and final file location. A failed render should return to a visible queue, not silently disappear or be marked complete. Retry only after you know whether the problem was an input, a temporary service issue, or a scene mismatch.
- Set a clear status such as queued, generated, needs-fix, approved, or released.
- Record the exact correction before a retry so repeated mistakes can be traced.
- Keep an approved output linked to the source SKU and campaign version.
- Open the final file and click its destination link before scheduling publication.
Catalog video workflow questions
Can a catalog feed be sent directly to any video generator?
Do not assume so. Confirm the provider’s current official import methods and supported fields, or prepare a reviewed export that matches its documented inputs.
Can generated text show an exact price?
Use a controlled text layer or a final edit for exact prices and offer dates, then verify each version against the current catalog.
Should every product use the same scene?
Only when the products fit the same visual setup and can be represented accurately. Different shapes, use cases, or safety requirements may need distinct scenes.
