The Wrong Unit of Measurement
Walk into any DTC growth meeting and you'll see roughly the same artifact: a screenshot of the Meta Ads Manager, sorted by spend or ROAS, with the team picking the top performers and saying "let's make more like that."
It's the most reasonable thing in the world. It's also the wrong unit of analysis.
The finished ad is what runs. But the finished ad is a composition — a hook + a demo + a proof moment + a CTA + an editor's pacing decision + a route to a specific audience. When the ad performs, one or more of those ingredients did the work. When the ad fails, one or more of those ingredients didn't.
If you don't know which, you're guessing every time you brief the next batch.
How "Make More Like This" Goes Wrong
A team sees a winner. They say "let's make more like this." They send the screenshot to the agency. The agency makes three variants. Same shoot location, similar voiceover, same product, similar pacing.
Two of the variants flop. The third works moderately well. The team assumes the variation that worked must be doing the same thing the original did — but actually, it might be a totally different ingredient earning the spend. Maybe the original ad won on the hook (first 3 seconds, pattern interrupt) and the new winner won on the proof moment (closeup of a satisfied customer at 0:18). The team has no way to tell.
Next cycle, the brief says "more variants of [winner]." The cycle perpetuates with no learning. Cost-per-acquisition holds steady or drifts up. The team blames "creative fatigue" and asks for more production budget.
The real problem isn't fatigue. The team has been compounding noise instead of signal — because the unit of analysis was the ad, not the ingredient.
What Ingredient-Level Analysis Looks Like
A creative supply chain treats every finished ad as a Bill of Materials. Each line item is one ingredient with attributable provenance:
- Hook (0:00–0:03): Creator A, UGC Reel #214, pattern-interrupt style
- Product shot (0:05–0:09): Asset 014, brand library, spring shoot
- Proof moment (0:12–0:18): Founder clip, recorded May 2026, conversational
- Demo (0:18–0:24): UGC asset, Creator B, kitchen environment
- CTA (0:26–0:30): Testimonial 23, closing line variation, urgency-styled
When the ad runs and performs (or doesn't), the system attributes the result back to those five ingredients. Over enough launches, patterns emerge: Creator A's hooks consistently win across categories. Founder clips drive spike in close-rate but flatten reach. UGC demos beat studio demos on cold audiences but lose to studio on retargeting. These are usable, structural insights — not "make more like this."
That kind of analysis isn't possible without two things: ingredient tagging at the asset level, and a system that connects performance data back through the recipe. Most brands have neither. Their analytics layer was built for the ad as the unit; their creative layer was built without ingredient-level tracking.
Why the Industry Defaults to the Wrong Unit
Three reasons.
Ad platforms surface ads. The Ads Manager interface shows you the ad as the row. ROAS is per-ad. Creative reports are per-ad. The default unit of analysis matches the default unit of the dashboard. It takes deliberate effort to break out of that frame.
Creative work has historically been bespoke. Until recently, ads weren't built from reusable ingredients. They were produced as one-offs. Reuse meant "use this ad again," not "extract the hook and reuse it in three new compositions." The ingredient frame requires that production output is itself indexed and addressable, which requires the supply-chain layer that most brands skip.
Attribution at the ingredient level is hard to demand. Most agencies and media buyers don't offer it. Brands don't know to ask for it. The result is a market equilibrium where nobody is doing the work, so nobody expects it, so nobody is doing the work.
What Breaks When You Switch Units
Three things break — usefully.
The "we need more creative" trap dissolves. When you can see that the winning ad was a recombination of three ingredients you already had, you stop asking for more production and start asking for better routing. The production budget often goes down. The activation rate goes up.
Creative reviews change. Instead of approving finished ads, the team reviews ingredients — and recipes (composition patterns). The conversation moves from "I like this one better" to "this hook archetype is earning, let's request three more from creator B."
Briefs get specific. A brief at the ingredient level looks like: "Three hooks, pattern-interrupt style, sub-3 seconds, against pain point A, from creators B or D." A brief at the finished-ad level looks like: "More lifestyle UGC like this." The first one ships better creative. The second one ships another guess.
How Marshal Operationalizes This
Marshal installs the naming layer that makes ingredients addressable in the first place. A creative gets an ID at the brief, the ID carries through the filename, the ad name, and the tracking link, and from that point a launch can be traced back to what it was made of.
That is Sprint 1, The Install: $15,000 total, billed $5,000 a month across three months or less, with a handover date named at kickoff. It does not judge which ad won. Your analytics tools already do that, and they keep doing it. What the install adds is the join, so the report can break out by ingredient instead of by screenshot, and the next brief is written against what actually ran.
That's the difference between running a creative operation and running a creative supply chain.
The unit isn't the ad. It's the ingredient. The ad is just where the ingredients happen to land for a few seconds before the spend tells you whether they worked.
If your creative reviews are still happening at the ad level instead of the ingredient level, the Creative Flow Review is a free 30-minute diagnostic. We'll show you what your last quarter's ads looked like at the BOM level — and what learnings were missing.