The problem
Picture your favourite meal at your favourite restaurant. You want to make it at home, so you look up the recipe and buy all the ingredients. You make it. It doesn’t taste as good as the restaurant.
The recipe wasn’t wrong. You just didn’t get the rest of it: the cook who has made it a thousand times, the equipment, the room you ate it in. That is what happens every time a creator copies a winning script.
Layer 1: the recipe. What the video says.
The hook. The claim. The problem it names. How it explains why the product works. The proof. The offer. The call to action. The order it all comes in. This is the recipe card. It is the easiest part to see and the easiest part to copy. You can read a script in a minute. That is why it is where almost everyone stops.
Layer 2: the cook and the plate. What the viewer sees while the words are said.
The creator. Their face. The room. Whether the product is in frame. The captions. The cuts. Screenshots, before-and-after photos, demos. And when each of those shows up compared to the words. Same recipe, different cook, different dish. Everyone knows this about food. Almost nobody applies it to video.
Take one line: “I thought this stuff was fake.”
- Said at a kitchen table with plain captions.
- Said while holding up a screenshot of hundreds of reviews.
- Said while holding an old photo of yourself, then cutting to now.
Same words. Three different videos. They will not perform the same. If all you saved was the script, all three look identical in your notes.
Layer 3: the kitchen. Everything around the video that it didn’t control.
The product. The price. The commission. How long the product has been on TikTok Shop. How much the brand is spending on GMV Max. How many other creators are already posting it. A great dish on a slow Tuesday in an empty restaurant still looks like a flop. A great video on a product the brand stopped pushing looks like a bad video.
Layer 4: the reviews. What the audience actually did, in numbers.
This is the layer where the data lives. Every metric TikTok gives you is a question the audience answered:
- 2s view. Did they stop scrolling?
- 6s view. Did they stay once they stopped?
- Average watch and completion. Did they hear the whole pitch?
- CTR. Did they want to see the product?
- CTOR. Once they saw it, did they buy?
- Comments. What did they push back on, ask, or repeat?
Most people read these as a report card. Views good, sales bad, move on. Read them as a timeline instead. The numbers tell you when people left. They don’t tell you why. A one-star review tells you the dish failed. It doesn’t tell you if it was the recipe, the cook, or the restaurant.
The example
Every number points back at one of the layers above it.
| What the numbers show | Layer to read | Why |
|---|---|---|
| Weak Hook | The cook | People swiped before the words mattered. It is the first frame, the first face, the first text on screen. Almost never the script. |
| Strong Hook, weak Payoff | Cook or recipe | They stopped, then left. The spoken hook didn't pay off, or the creator started talking before anything interesting appeared on screen. |
| Strong Body, weak Click | The recipe | They watched but didn't want the product. No clear reason to look, no mechanism, no proof, or the pitch never landed on a real problem. |
| Strong Click, weak Convert | The kitchen first | They wanted it and didn't buy. Price, reviews on the product page, whether the brand is still pushing spend. |
| Same drop across all creators | The kitchen | Stop reading the video. |
| Drop on one creator only | Cook or recipe | Now read the video. |
A number alone can’t tell you which one. A number plus what was on screen plus what was being said at that second can. Here is what a full read of one video looks like:
| Time | Said | Shown | What the audience did |
|---|---|---|---|
| 0 to 2s | “Doctors don't tell you this” | Creator face, bold caption | Most still here |
| 2 to 6s | Opens the question | Greenscreen, article screenshot | A third leave |
| 6 to 11s | Names the problem | Before photo | Holds steady |
| 11 to 16s | How the product works | Product in hand | Clicks start |
| 16 to 23s | Reviews and proof | Review screenshots | Clicks speed up |
| 23 to 30s | Price and CTA | Product plus price | Orders land here |
Now you aren’t asking “why did this video win?” You’re asking “what was being said and shown at the exact second the audience changed its mind?” That question has a reusable answer. The first one doesn’t.
This is the job
Most creative strategists do one thing: break down scripts. Pull the transcript, find the hook, name the angle, rewrite it for their brand, send it to creators. That is reading the recipe. It is a start. It is not strategy.
Strategy is holding all four layers at once. The words, the screen, the kitchen, and the numbers, lined up in time. Then changing one thing and checking whether the numbers moved.
The honest problem is that this is easy for one video and nearly impossible for a hundred. One video: pull the transcript, watch it twice, note what’s on screen, line it up with the retention curve. An hour, maybe two. Now do it for every video from fourteen creators on six products over a month, and remember which of last quarter’s tests already failed. Nobody does that by hand. So they fall back to the recipe, because it is the only layer that fits in a spreadsheet. That is the gap. Not knowledge. Capacity.
Once you judge videos by layer, you stop collecting recipes and start collecting rules. Creator face holds viewers better than product B-roll in the first eight seconds. Before-and-after proof drives the click, not the claim. The price reveal costs retention but raises orders. A demo beats a spoken claim for the same mechanism. Rules outlive the dish. When the product fades, the rule moves to the next one. That only works if you write down which layer killed a video, not just that it died. A list of failures is not a list of lessons.
Do this in Verdict
Verdict is built around the four layers, so the read above is something you do every day instead of once.
- 1The reviews. The Leak Finder shows every video’s funnel in one row, Hook through Convert, and marks the leak. It compares the video against the ones that converted on the same product in the same window.
- 2The kitchen. The Pre-Slot Screener and its Window Score show you the kitchen before you cook. Inside the Leak Finder, the banner above the list tells you when something moved for the whole brand in that window. That is the “same drop across all creators” read, done for you.
- 3The cook. Video Breakdown returns the script and a timestamped list of what is on screen, together. That is the Said and Shown columns above, without watching the video twice. The content style tag in the Creative Matrix is the quick version, and the Matrix puts style, hook, angle and account side by side across every video.
- 4The recipe. Waves shows which lines show up in more than one winning script. Scripts groups scripts that say the same thing in nearly the same words. The Element Library breaks winners into hooks, bodies and CTAs. Script Lab is where you write your version.
- 5The rules. Every verdict goes into Learnings. When a video dies, note which layer killed it. That is the memory the next version runs on, and the part no one keeps in their head.