Batching a Month of Brand Video in One Afternoon

A month of brand video — call it 24 short-form assets plus four longer cuts — used to mean a two-day shoot, a week of editing, and a production budget with a comma in it. The version I'm going to describe takes an afternoon, roughly four and a half hours, and one person who knows what they want. It is not magic and it is not free: the yield rate is around 70%, the setup work is real, and there are formats it simply doesn't cover.

The mental model that makes batching work is a factory floor, not a to-do list. You are not making 24 videos. You are running five stations in sequence, each of which processes all 24 assets before the next one starts. The efficiency comes from never re-loading context — when you're generating product motion, you generate all the product motion, with the same references, the same model, the same settings.

Do it the other way — one video start to finish, then the next — and the same output takes three days. I've measured it. The difference is entirely context switching.

The setup work that happens the week before

Batch day is execution only. Four things must exist before you start:

A script bank. All 24 scripts written and approved. This is the single biggest predictor of whether batch day works. Writing scripts during batch day is how an afternoon becomes two days.

A locked style spine. One decision set applied to everything: aspect ratios (9:16 primary, 16:9 for two of the longer cuts), color direction, caption preset, voice, music family. Write it down as a one-page document. Anything not on that page gets decided once, on the fly, and then added to the page.

Reference assets. Product stills from three angles, presenter reference images if you're using one, environment references if your brand has a look. These are what buy you consistency across 24 separate generations — the difference between a coherent month of content and 24 videos that look like they came from 24 companies.

A shot inventory. Read all 24 scripts and list every distinct shot needed. You'll usually find the 24 videos need about 45 unique shots, not 90, because scripts repeat setups. That deduplication alone saves an hour.

The five stations

Station	Time	Output	Yield you should expect
1. Anchor stills	45 min	20–30 reference frames	60% keep rate
2. Motion	90 min	45–60 clips	70% keep rate
3. Voice	30 min	24 voiceovers	90%+
4. Assembly	75 min	24 assembled cuts	95%
5. Ship	30 min	Captions, scheduling	100%

Station 1 — anchor stills. Generate the key frames first: product hero, presenter, each distinct environment. Everything downstream inherits from these, so a bad anchor propagates. Spend the time here; it's the cheapest place to fix a look. Use text-to-image and pick two or three keepers per setup.

Station 2 — motion. The long block. Feed anchors into image-to-video for controlled shots, and use reference-to-video when the same product or character has to appear across many clips — Seedance 2.0 Fast reference-to-video and Wan 2.7 reference-to-video are the workhorses here. Queue in waves: submit ten, work on prompts for the next ten while they run, review in batches of ten. Never wait on a single render.

Station 3 — voice. All 24 scripts, one voice, one settings preset, straight through. Half an hour. The reason this is so fast is that revision is free — a wrong line gets regenerated in seconds rather than re-booked.

Station 4 — assembly. Captions, overlays, trims, music. Treat these as batch operations rather than per-video creative decisions: apply the caption preset to all 24, then review all 24, then fix the handful that broke. Reviewing one at a time triples the block.

Station 5 — ship. Caption copy, platform selection, scheduling across the month. Space the publishing dates now, while you have the whole month in front of you and can see the sequencing.

Yield: the number nobody quotes honestly

You will not keep everything you generate. Realistic keep rates from actual batch days:

• Anchor stills: about 60%. You generate three, keep one or two.
• Motion clips: about 70% on the first pass with good anchors, closer to 45% without them.
• Voice: over 90%, since scripts were already approved.
• Assembly: near-total, because the creative decisions were made upstream.

Multiply through and a 24-video month costs roughly 55–70 video generations plus 40–50 image generations. Budget credits on that basis, not on a one-generation-per-asset fantasy. Forecast your monthly credit spend on that basis and it stops being a surprise.

The lever that most improves yield is anchor quality. Teams with a weak Station 1 spend Station 2 regenerating, and the afternoon becomes an evening.

Keeping 24 videos looking like one brand

This is where batching earns its reputation. Coherence is a byproduct of process, not of taste.

• Same references everywhere. One product reference set, one presenter set, reused across every clip that features them.
• Same model per shot type. Don't mix four video models across a month's b-roll — each has its own motion signature and grade, and mixed sources read as sloppy even when each clip is individually good.
• Same caption preset and safe margins. Applied at assembly, never per-video.
• A pinned first frame for recurring formats. If your weekly explainer always opens the same way, generate that opening once and reuse it.

A written spine document handles most of this; a one-page spine document covers what belongs in one.

What batching doesn't cover

Three categories should stay outside batch day:

Timely and reactive content. Anything responding to news, a trend, or a competitor's move needs to be made in the moment. Leave 20% of your monthly slots empty for it — a fully pre-booked calendar can't react.

Anything requiring a person's real presence. Announcements, apologies, genuinely personal moments. Film those.

Multi-scene narrative pieces. A three-minute story with continuity across eight scenes isn't a batch item; it needs its own session with scene chaining and per-scene review. Movie mode exists for exactly this, and it's a different rhythm.

Making the second batch day faster than the first

The first batch day will overrun. Expect five to six hours and about 18 usable assets. Three things compress it afterwards:

1. Reuse the prompt shelf. Every prompt that produced a keeper gets saved by shot type. By month three you're retrieving rather than composing, and Station 2 drops by 30 minutes.
2. Turn recurring formats into workflows. Any format appearing four or more times in a month should be a saved, reusable workflow that runs with new inputs and can auto-post on a schedule. See Versely's workflow library for starting points.
3. Repurpose deliberately. A month's batch produces raw material for far more than 24 posts — cutdowns, stills, carousels, quote cards. Batching 30 days in one sitting is the companion read on cadence and repurposing.

FAQ

How many videos can one person realistically batch in an afternoon?

Twenty to twenty-eight short-form assets with scripts pre-approved and references ready, at about 70% first-pass yield. Without pre-approved scripts, expect eight to twelve — script-writing is the bottleneck, and it doesn't belong in the same block as production.

Won't a whole month made in one sitting look repetitive?

It can, and the fix is deliberate variety at the script stage rather than the production stage. Aim for at least four distinct formats across the month and rotate the opening structure. Visual consistency is desirable; structural sameness is not.

What's the biggest cause of a batch day running over?

Weak anchor images. If Station 1 output is mediocre, every clip in Station 2 needs extra attempts, and a 90-minute block becomes three hours. When anchors aren't landing, stop and fix the prompt rather than pushing forward into motion.

Should we batch across multiple campaigns at once?

Yes — batch by production type, not by campaign. All anchor stills for every campaign, then all motion, then all voice. Campaign-by-campaign batching reintroduces exactly the context switching you're trying to remove.

How far ahead should the calendar be batched?

One month is the sweet spot. Two months of pre-made content tends to age badly against product changes and market shifts, and the last two weeks of it usually get rewritten anyway. Keep a fifth of the slots open for reactive work.

Pick a month you actually need to fill, write the scripts this week, and run one batch day with the five stations in order. If your format repeats, save it as a workflow afterwards so the next month costs half as much attention.