Social media video production, day by day for ten days
Social media video production for paid media is a scheduling problem as much as a creative one. The work has to arrive before the current ad fades, which means the calendar is fixed and the concepts fill it, not the other way round.
Most production timelines are built around a shoot. Paid social does not work that way: the constraint is not the camera, it is that Meta consumes creative faster than a traditional process replaces it.
Here is the ten day cycle we run, day by day, including where the time actually goes and the two days that get wasted most often.
Why the calendar is fixed and the work fills it
A project-based process starts when someone asks for something. By then the ad that needed replacing has been fading for a week, and the request arrives as an emergency.
A fixed cycle inverts that. The delivery date exists before the concepts do, so the account always has something new to test whether or not anything broke. It is a loop rather than a project, which is the reason it keeps working after the first batch.
Days 1 and 2: read the account, not the brief
Production starts by looking at what is already running. Which ads are fatiguing, where hold rate collapses, which angle has never been tested, what the audience has now seen too many times.
This is the step most studios skip, because it is not production. It is also the step that decides whether the next eight days produce anything useful.
- Ad level metrics, not campaign averages, so a single fatiguing winner is visible.
- Hook rate and hold rate first, cost second.
- What has been tried before, so the cycle does not repeat a losing angle in a new costume.
Days 3 and 4: the hypothesis and the scripts
Every concept has to answer one question before anything is built: what is this testing, and why do we think it beats what is running now?
If that cannot be said in a sentence, the concept is a refresh rather than a test, and it will produce a result nobody can act on. Scripts and storyboards come after the hypothesis, not before.
The format is chosen here, and it follows the diagnosis. Claymation for ad blindness, a podcast slice for a trust gap, a demonstration when the product needs explaining before it can be wanted.
Days 5 to 8: the build
Production runs in parallel rather than in sequence, which is what makes ten days possible. Claymation, 3D, talking objects and creator-led cuts are all in progress at once, from assets that already exist rather than waiting on a shoot.
- Each concept becomes several variants across hooks, angles and openings.
- Everything is cut for sound off first, with captions inside the safe area.
- Placement ratios are produced natively rather than cropped from one master.
One idea producing several genuinely different variants is what turns a delivery into a test. One file is not a test, it is an opinion.
Days 9 and 10: delivery that can be read back
Ads arrive ready to upload in 9:16, 4:5 and 1:1, named and tagged so that when a result comes in, it is obvious which hypothesis produced it.
Delivery on a known date is the part clients notice most. The media buyer plans budget around a drop instead of chasing assets, which is usually the first thing that changes when a team moves to a cycle.
The two days most studios waste
First: waiting for a shoot. A shoot is a fine way to make one thing and a terrible way to make twelve. Most of what performs on paid social can be built from assets that already exist.
Second: revisions on taste. Rounds spent on colour, pace and music feel like progress and change nothing about whether the ad stops the scroll. Revisions are worth spending on the hypothesis and the first three seconds, and almost nowhere else.
What one cycle actually contains
Volume follows spend, because a bigger account exhausts creative faster:
- 2 core concepts and 7 total ads per cycle for accounts spending $10k to $30k a month.
- 5 core concepts and 19 net-new variations, plus a mid-month resprint, for $30k to $100k.
- 8 concepts and 32 deliverables across markets for $100k and above.
The resprint matters more than it sounds. Cutting the data mid-month and rebuilding whatever is pulling is how a cycle compounds instead of restarting every ten days.
Questions we get asked
- How long does social media video production take?
- We run a fixed ten day cycle from reading the account to delivering finished ads. The length matters less than the fact that it repeats, so a tested replacement exists before the current winner fades.
- Do you need a shoot for social media video ads?
- Usually not. Claymation, 3D, talking objects, motion and editorial formats are built from assets that already exist, which is what makes a ten day turnaround possible. Shoots make sense for one hero asset, not for volume.
- How many videos do you need per month for Meta?
- Enough that the next test is ready before the current one fades. In practice that means a new set of concepts every one to two weeks, with the number of concepts rising with spend rather than the number of files.
- What is the difference between video production and ad creative production?
- Video production ends at a finished film. Ad creative production starts from a hypothesis about the account and ends with several variants built to be compared against each other in the platform.
- Who writes the scripts?
- We do, after reading the account. The script follows the hypothesis, which is why the account read happens before anything is written rather than after a brief arrives.