Before production starts (tips 1 to 5)

These five decisions set the size and shape of everything that follows. Each one is cheap to make now and expensive to change once files exist.

1. Count the assets your plan implies before you promise a launch date

Owner: Creative operations, with ecommerce and paid media, 10 to 12 weeks before your first retail event

Peak-season plans often get approved as a slide with four campaign concepts on it. Teams often don't multiply out what those concepts mean in files until production has started, and by then the launch dates are public.

The retailer calendars alone show how fast the number grows. Target's weekly Sunday deals in 2025 meant eight separate deal cycles, each needing its own set of assets. Walmart's three events meant three offer states, and Amazon's Black Friday Week needed creative that held up for eleven days. That's before you factor in sizes and markets. Take one approved concept and walk it through every dimension that varies in your business.

That's one concept. A portfolio running four concepts across three product lines lands in the thousands. Your rows will differ (Target doesn't sell in the UK, for one), but the exercise is the point: run it on your own plan and put the answer next to what your team and agencies can deliver before the first go-live date. When markets are part of the count, creative localization adds its own review time on top.

The count also tells you where to put effort. Find the multiplier creating the most volume and systemize that layer first. Unilever's volume came from products, so it built one 3D digital twin per product for Dove, TRESemmé, and Vaseline, holding every pack format and language label in one file. Product imagery now comes out twice as fast at half the cost. If your volume comes from offers and waves instead, the offer layer is the one to systemize, which starts with the discount.

2. Set discount guardrails before the first file carries a number

Owner: Ecommerce, with finance, before any offer is locked

When discount depth moves late, every file carrying a percentage moves with it. That's the production reason to settle depth early, and to settle it as a range with a floor and a ceiling rather than a number that gets revisited every week in October.

Two data points are worth bringing to that conversation. Salesforce found Cyber Week 2025 average selling prices rose 6% while order volume grew only 2% globally, so much of last season's record was price rather than units. Returns took another slice: $181 billion of online holiday purchases, 14% of everything bought, came back (Salesforce). Deeper discounts don't guarantee the volume, and some of what they win comes back.

Brands handle this differently. MAC Cosmetics kept 30% off sitewide, in line with the year before, and put its effort into TikTok Shop livestreams. The Ordinary ran "Slowvember", 23% off for five weeks on its own sites only, so retail partners weren't undercut. Either way, the decision was made once, and the creative could be built around it.

3. Lock offers by retailer before a designer opens a file

Owner: Ecommerce, signed off by brand and legal, 8 to 10 weeks out

A settled discount is still one number. Each retailer turns it into a different offer, on a different date. A concept that changes costs a round of design. An offer that changes after production starts invalidates every file carrying a price, a percentage, or a date. You're rebuilding, not revising, in the only weeks of the year when nobody has slack. Settle these before design starts:

• Products in and out: With a named owner for exceptions

• Discount per category: Within the guardrails from tip 2

• Wave structure: What goes live in October, in each retailer's Black Friday wave, and in the final hours

• Retailer-specific variations: Where a retailer needs different pricing, claims, or dates

• Claims and legal copy: Approved in writing before it goes into a layout

Then map those decisions to each retailer's calendar, because no two match.

Retailers can also move a price for you. Amazon's ad policy requires creative to match the live promotion, so when a deal changes, the ads have to change too. How often that happens is harder to pin down, but in one Seller Central discussion, a seller described an approved 30% deal needing 35%, then 39%, to keep its badge (Amazon Seller Central). Decide in advance what happens to the assets when an offer changes.

How ScarAway did it: ScarAway, an Alliance Pharma brand, sells through Amazon, Walmart, CVS, Meijer, and its own site. Working from agency-made masters in Rocketium's AI Studio, the team produced 137 product pages, A+, and video assets and cut brief-to-first-draft time to under 24 hours, down from about five days.

• Do the same: Store every offer as fields (price, percentage, dates, retailer) and build the files from those fields, so an offer change regenerates assets instead of reopening them.

• Do it differently: Write a separate offer row for each retailer even when the discount matches. Dates and badge wording rarely do, and the row is what your creative brief should point to.

4. Map approvals before the assets exist

Owner: Creative operations, with brand, legal, and retailer account leads, about 8 weeks out

Locked offers tell you what the assets will say. The approval path decides how fast they go live saying it. Approval is a throughput limit, but many production plans still treat it as a final review step.

Here's how that plays out. A team produces two thousand assets in a week. Brand review is one person with a full calendar, legal wants every pricing claim, and the first retailer rejections take four days to come back. The files were on time. The launch wasn't.

• Named approvers per asset type: With a backup for each

• A written checklist: Review becomes pass or fail against criteria, not opinion

• A rejection buffer: How many days you can absorb, and what happens on the fourth

• Checks before submission: Retailer rules, brand rules, and legal claims

Retailer rules deserve their own line on that checklist. Amazon asks advertisers to submit ads at least a week before launch, allows up to 72 hours to review translated campaigns, and prohibits savings claims like "Save 50%" and superlatives like "Huge savings" in ad copy. Black Friday copy that works on your own site can fail on Amazon without anyone changing a word.

Mondelez splits the work in a way worth borrowing. Its generative content tool checks against a list of banned content, and every asset still gets human review before it publishes. Automated checks for size, safe zones, and banned words first, then people for the judgment calls. Our guide to brand compliance covers what to automate and what to keep human.

5. Build masters that adapt

Owner: Creative lead or agency, 6 to 8 weeks out

With offers fixed and reviewers named, the last decision before production is the file everything else is built from. A master that looks beautiful as a hero image and falls apart at 160 by 600 pixels is still a mood board. The real test is how many adaptations it produces before someone has to rebuild it by hand. A master that passes has these five things:

• Layered and editable: No flattened artwork, no copy baked into images

• Reflowable text: Headlines can lose a line without breaking the layout

• Crop-safe focal points: The product stays readable in a square, a vertical, and a skyscraper

• Separate offer fields: Price, percentage, and dates live in their own layers

• Defined logo and legal zones: Compliance elements don't get cropped out at small sizes

Get this right and adaptation becomes mechanical, which means it can run in volume. Get it wrong, and every size is a small design project, which is how a four-week production window turns into seven, and where the hidden cost of manual versioning builds up. Coca-Cola's 2025 holiday campaign is a good model: it started as a 60-second film and a 30-second spot, then ran across outdoor, digital, social, in-store, on-pack, and its truck tour in three regions. A useful brief test is to design with the smallest placement in mind. If the idea survives a sponsored product tile, it will survive a billboard.