A catalogue with four hundred products and budget for forty videos is a prioritisation problem, and most brands solve it with a spreadsheet sorted by revenue.
That is a reasonable start and a poor finish. Revenue tells you what sold last quarter under the current creative. It does not tell you where a video would change the outcome.
Here is the ranking I would use instead.
The five filters
Score each SKU, then sort. The weights are mine; adjust them, but keep all five.
| Filter | Why it matters | Weight |
|---|---|---|
| Traffic already arriving | Video improves conversion of existing visitors, not of traffic you do not have | 30% |
| Margin per unit | A video on a 12% margin product has almost no room to pay for itself | 25% |
| Weakness of current creative | The bigger the gap, the more a video can move | 20% |
| Return or support rate | Video is the cheapest way to answer "does it work like that?" | 15% |
| Paid support behind it | Video only compounds where spend is running | 10% |
The output is a list that looks different from a revenue sort, and it should. The top of my list is almost never the hero product — the hero already has good creative, an established page and a customer who is not asking questions.
Why traffic first, not revenue
Video does not create demand. It converts attention you already have.
A SKU with 8,000 monthly visits and a 1.4% conversion rate has 112 orders and an obvious question: what would 2.0% look like? That is 160 orders from the same traffic, and a video is a plausible mechanism.
A SKU with 300 visits and a 3% conversion rate converts better and is worth less, because the ceiling is the traffic. Video cannot fix a distribution problem.
The three SKUs to skip
The bestseller with great creative. It is already working. Spending your first video on it buys you nothing measurable and sets a precedent that video goes to the safe choice.
The long-tail single-unit item. Extra cost per unit, no repeat potential, and a page nobody visits. A still is sufficient.
The product with a fulfilment problem. If the review page is full of arrived late or wrong colour, video will not help and the data will be polluted by a problem the video cannot touch. Fix the operations first, then measure creative.
Batch size: ten to twenty
I have seen two failure modes, and they are symmetric.
Too small. Three videos and no pattern. Whatever happens is noise, and you will attribute it to the video because that is what you changed.
Too big. Two hundred videos, no control group, no way to know which of forty variables moved the number. Expensive and uninformative.
Ten to twenty SKUs, chosen to include at least three that you expect to do nothing, gives you a comparison and a surprise. The controls are the point: if everything improves, you are measuring seasonality.
What to actually measure
Not "did sales go up". Four numbers, before and after, per SKU:
1. Conversion rate on the PDP — the metric video is aimed at.
2. Add-to-cart rate — distinguishes "did not understand" from "did not want".
3. Return rate at 30 days — the comprehension effect, which shows up later and is usually bigger.
4. Cost per acquisition on the paid variant, if any — for the SKUs with spend behind them.
Expect the return-rate change to be the surprise. A demo that shows the actual size of a bag or the actual texture of a fabric prevents a purchase the customer later reverses.
The format decisions, made once
Once the SKU list is set, the production spec is the same for all of them. Decide it up front:
| Decision | Default |
|---|---|
| Ratio | 9:16 master, 1:1 and 16:9 cut from it |
| Length | 6–10 seconds for PDP, 15–20 for paid |
| Sound | Captions burned in, works muted |
| Product reveal | First frame, not second eight |
| Claim | One, and only one that survives legal review |
| Files | Master plus per-platform exports, named by SKU |
Naming by SKU sounds trivial and saves an afternoon every month. A folder of final_v3_edited_FINAL.mp4 is a catalogue rollout that cannot be maintained.
The sequencing that avoids rework
Ship in three waves, not one:
Wave 1 — ten SKUs. Learn what your audience responds to. Ideally include two languages if you sell in more than one, because that decision is cheaper to make at ten than at two hundred.
Wave 2 — the next twenty. Apply what wave one taught you. This is where the cost per video should fall, because the template is established.
Wave 3 — the remainder. By now you know which categories video does not help, and you skip them deliberately rather than discovering it expensively.
Where NNI sits
Catalogue work is what the marginal ladder is for. From $70 a finished video, $65 each from the twentieth, $60 from the fortieth — the per-asset price falls precisely because a catalogue is repetitive production, and repetitive production is cheaper per unit.
For a first wave, order ten and look at the data before you order eighty. The ladder will still be there, and a wave that starts with evidence is a wave you can defend.
---
Jayesh runs NNI Design Studio, white-label video production for agencies and brands. Rate card at nnidesignstudio.com/pricing; the brief template we work from is a free download on /resources.
Drop your take 0
// no hot takes yet. start the thread 👇