How to ship 400 catalog ads in a day (and why your agency can't)
A field guide for ecommerce teams: how to turn a product feed into on-brand catalog ads at ad-ready resolution, without writing a brief for every SKU.
If you run performance at an ecommerce brand, you know the math. Every extra creative concept in the account is worth measurable ROAS. Meta, TikTok, Pinterest — they all reward variance, and they all punish creative fatigue. The winning teams ship dozens of variants a week. Most teams ship one.
The blocker is not talent. It's the workflow.
The old workflow (and why it dies at scale)
- Merch picks 40 SKUs for the drop on Monday.
- Marketing writes a brief for the hero SKU on Tuesday.
- Design mocks the hero in Figma on Wednesday.
- Legal comments in a doc on Thursday.
- The agency resizes the hero into 8 ratios on Friday.
- Everything else gets a boosted product photo with a purple gradient.
That workflow was built for a world where you shipped 3 hero campaigns a quarter. It cannot survive a world where you ship 3 drops a week. Notice what it optimises for: the quality of one asset. Notice what it never touches: the other 39 SKUs, which together are most of your revenue.
The new workflow
- Encode your brand once — type, palette, ribbon rules, disclaimers, illegal claims.
- Point at your catalog — Shopify, Centra, a CSV, whatever.
- Design one template — hero moment, offer stack, CTA.
- Apply the template across every SKU in one pass — Xeli renders 400 on-brand cards, sized for every channel, in about the time your kettle takes.
- Approve as data — reviewers see one queue, comments live on the artboard, versions are diffable.
- Ship — sizes and metadata land in the channel spec each platform wants, and performance flows back to the SKU.
| Step | Old workflow | Templated workflow |
|---|---|---|
| Brief | Per campaign, per SKU | Once per concept |
| Design | One artboard per asset | One template per concept |
| Resize | Manual, per placement | Rendered per placement |
| Review | Doc comments and Slack | Queue with versioned state |
| Coverage | Hero SKUs only | Full catalog |
| Time to 400 assets | Weeks | A single run |
A worked example: a 40-SKU drop across four placements
- Step 1 — Feed check. Confirm every SKU has a clean cutout or lifestyle image, a correct price, a variant label, and inventory. Skip nothing here: feed errors multiply by the number of assets you are about to render.
- Step 2 — Brand rules. Lock the type ramp, the palette, logo clear-space, badge treatment for offers, and the disclaimer text per market. This is the artefact you reuse for the next year.
- Step 3 — One template, three concepts. Build a product-hero layout, an offer-led layout, and a social-proof layout. Three concepts is enough to learn from and few enough to keep consistent.
- Step 4 — Render. 40 SKUs × 3 concepts × 4 placements is 480 assets. Rendered from templates, each already sized and safe-zone-correct for its placement.
- Step 5 — Review the outliers, not the set. Spot-check by rule: any SKU with missing imagery, any price that looks wrong, any text that overflows. Approve the rest in bulk.
- Step 6 — Ship in structured batches. Launch one concept per ad set so results roll up cleanly, and hold the third concept back as the refresh for week two.
The three habits of high-velocity creative teams
1. They treat the template as a product. It has versions, owners, and changelogs. Every campaign is a fork of the master, not a from-scratch brief.
2. They ship variance on purpose. Instead of debating one hero, they ship 12 executions of the same message and let the platform algorithms pick the winner.
3. They read the numbers back to the catalog. Which SKUs converted at which price on which ratio? That becomes next Monday's merch call.
What to measure so the loop actually closes
- Hook rate — the share of impressions that hold attention past the first moments. Diagnoses the opening, not the offer.
- Hold rate — retention deeper into the asset. Separates an interesting hook from a persuasive one.
- Cost per result by concept, not by asset — asset-level data is too thin to act on.
- Frequency by audience — the earliest reliable signal that a refresh is due.
- Shipped variants per week — the leading indicator for everything above.
Why your agency can't do this
Not because they're not talented. Because the unit economics don't work. An agency bills for the brief. Nobody writes a brief for the 273rd SKU in a catalog. Which is exactly the SKU that is going to blow out on TikTok this Thursday.
The agency model is excellent at the thing it is priced for: a small number of high-craft concepts. Pair it with a system that executes those concepts across the catalog, and both halves work. Ask it to do both and you get hero ads plus a long tail of gradients.
That is the gap Xeli exists to close: 400 shelf-ready catalog ads, on brand, ready to ship, before the standup ends. Stores under 10 SKUs can run the whole loop free on the Launch plan.
Key takeaways
- →Clean the product feed before you render anything — errors multiply by asset count.
- →Write the brand rules down once; treat that document as the highest-leverage asset you own.
- →Build three concepts per drop, not one hero, and hold one back as the week-two refresh.
- →Review by exception with rule-based spot checks instead of asset-by-asset approval.
- →Track shipped variants per week as a first-class performance metric.
Frequently asked questions
- How long does it actually take to render 400 catalog ads?
- The rendering itself is minutes once templates and the feed are ready. The real work is upfront: cleaning the feed and encoding brand rules. Teams typically spend a day or two on that once, then reuse it for every subsequent drop.
- Won't 400 templated ads look repetitive?
- They look consistent, which is different. Variance should come from concepts, hooks and offers — three to five distinct layouts per drop — not from each asset being individually designed.
- Do I still need an agency?
- For high-craft concepts, brand films and campaign ideas, agencies remain valuable. What does not fit their pricing model is the long tail of catalog executions, and that is exactly the part a system should own.
- What do I need before I start?
- A product feed with clean imagery and accurate prices, a defined brand system, and one placement you care about most. Everything else can be added later.
- Is there a free way to try this?
- Yes. Xeli's Launch plan is free for stores with under 10 SKUs, which is enough to run the full template-to-placement workflow end to end.
References
- [1]Meta Ads Guide — placement specs — Ratio, duration and safe-area requirements per placement.
- [2]TikTok Business Help — creative best practices — Guidance on creative variety and refresh cadence.
- [3]Google Merchant Center — product data specification — Feed field requirements referenced in the feed-check step.