Ad Creative Production for Ecommerce: The Best Workflow for Brands Running Catalog Ads

Ad creative production for an ecommerce catalog is the process of getting a usable ad image and ad copy attached to every product you advertise, not just the handful you feature. The workflow that scales has four steps: decide which products get custom creative and which run on a template, generate the images from product data rather than from a design brief, host them at versioned URLs so refreshes actually ship, and point your product feed at them. The failure mode is almost never taste. It is that a design queue priced per asset makes covering 900 SKUs irrational, so nobody does it, and the catalog serves whatever the store exported.

Why catalog creative is the part everybody skips

Look at where creative effort goes in a typical ecommerce ad account and the distribution is strange. A brand will spend two weeks and real money on a launch campaign built around four products, then run a catalog campaign covering the entire range on images that came out of the product information system with no involvement from anyone in marketing. The catalog campaign frequently serves more impressions than the launch. It is just that nobody sees it, because a catalog ad has no single finished version to review. The platform assembles a different one per person, so there is no artboard anybody signs off.

The economics reinforce it. If creative is bought per asset, whether from an agency, a freelancer or a credit-based tool, then covering 900 products has a price tag that has to be justified against the revenue of the long tail, and it never is. So the tail runs plain. Meanwhile the competitor whose feed points at branded compositions with a legible price is winning the same auction with a more persuasive frame at the same bid.

How do fashion brands create unique ad visuals for 1,000+ products without a big design team?

They stop treating each product as a design job and treat the brand as a system instead. One set of rules covers color, type, logo placement, badge style and crop, and every product image is generated against those rules from the product data itself. A designer builds and owns the system, which takes days, not the 1,000 outputs, which take minutes. That is the only structure that survives catalog scale.

This is also why the question keeps coming up in fashion specifically. Apparel catalogs have deep variant trees, a single style can carry a dozen colorways, and seasonality forces a refresh two to four times a year. A per asset workflow multiplies all three problems together. A systematic one absorbs them, because adding a colorway is new input into an existing system rather than a new brief.

Which products actually deserve custom creative?

Not all of them, and pretending otherwise is how these projects stall. Sort the catalog by ad spend over the last 90 days and the curve will be steep. A workable split looks like this.

The point of the split is that the long tail does not need to be good, it needs to stop being bad. Moving a tail product from a raw pack shot to a branded, correctly cropped, price-legible frame is most of the available gain. Squeezing a second round of polish out of it is not worth anyone's afternoon.

The four step production workflow

Once the tiers are set, the mechanics are the same for every ecommerce brand regardless of platform.

1. Get clean product data out of your own site. You need a title, price, image, availability and a landing URL per product, and they need to be current. Most stores can export this, but if your product information lives across a CMS, a PIM and a set of landing pages, it is often faster to pull structured product data straight off your own pages than to reconcile three systems by hand.
2. Generate the creative from that data. One brief per tier, not one per SKU. Every output should come back in your own colors, type and logo, in the crops the platforms serve.
3. Host at versioned URLs. This is the step that gets skipped and it is the one that breaks refreshes. More on it below.
4. Point the feed at the new images. Update image_link, re-upload or re-sync the feed, and confirm the platform picked the change up rather than assuming it did.

Step two is where a flat priced tool changes the arithmetic. If generating an image costs a credit, the long tail stays uncovered because covering it costs real money for uncertain return. If the price is the same whether you make ten images or a thousand, the calculation disappears and covering the whole catalog becomes the obvious default. That is the practical reason our own pricing is flat monthly with no per asset credits.

What image sizes does an ecommerce catalog feed need?

Meta requires product images of at least 500 x 500 pixels in the image_link field, as JPEG or PNG up to 8 MB, and TikTok states the same 500 x 500 minimum for square carousel placements. Both are acceptance floors rather than recommendations. Build at 1080 x 1080 or larger, because the placements these images serve into run well above 500 pixels and an upscaled source reads as soft.

Pinterest is the one that differs. Its shopping ads follow the standard image ad specs rather than a separate catalog spec, which means 1000 x 1500 at 2:3 is what the surface is built for, with a 20 MB file cap on desktop and 32 MB in app. If Pinterest is a real channel for you rather than a checkbox, the vertical version is worth producing rather than pushing a square through. The full field-by-field breakdown, including the feed text limits that differ from ad copy limits, is in our catalog ads specs and creative page.

How do you refresh catalog creative without breaking delivery?

Change the filename, not just the file. Meta's catalog documentation states that if you change the image later, the new image must use a different URL or the change will not be recognized. The platform caches the image against the URL string, so overwriting product-1234.jpg in place leaves the old creative serving indefinitely.

This is the most common reason a catalog creative refresh appears to do nothing. A team replaces images across an entire feed, watches performance stay flat, and concludes that catalog creative does not move the needle, when the new creative never actually shipped. Version the filename at the point the feed is generated, so product-1234-v2.jpg, and make it automatic rather than something a person remembers. Then spot check three live image_link URLs against the product pages before you draw any conclusion about the test.

What about the ad copy?

In a catalog ad the headline comes from the title field in your feed, not from a headline field in Ads Manager. Meta accepts up to 200 characters there but recommends 65 to avoid truncation, and most ecommerce exports produce titles built for a category page, stuffed with attributes, that blow straight past it. Rewriting feed titles to sit inside 65 characters with the brand and the product benefit first is one of the cheapest wins available on a catalog campaign, and it applies to every impression the feed will ever serve.

For the campaigns where you do control the copy fields directly, the constraints are different again and worth getting right per platform. Our ecommerce ad creative generator page covers the wider build, and the product ad maker goes deeper on single product creative.

How much should ecommerce ad creative production cost?

Judge it per covered product rather than per asset, because per asset pricing is what makes catalog coverage look unaffordable in the first place. A brand paying a per creative rate will rationally cover the hero products and leave the rest dark. A brand on a flat rate covers everything and spends its thinking time on the tier split instead of on approvals. We break the market rates down in how much ad creative costs.

The second cost nobody prices is latency. A design queue that turns creative around in a week sets the maximum speed at which you can respond to a stockout, a competitor promotion or a seasonal shift, and on a catalog that changes weekly, that latency costs more than the invoice does.

Start with the tier split

If you take one thing from this: open your ad account, sort products by spend, and draw the line between the products a human should think about and the products a system should handle. Almost every ecommerce brand discovers the line is further up than they assumed, and that the majority of their catalog has never had a single creative decision made about it. That is not a failure of taste. It is a workflow that was priced to make the decision for you.

Paste a product URL into the studio at the top of this page and you will get the images and copy for that product in every placement size, on your brand. Then do it for the tier that has been running plain.