A video production workflow that keeps costs down and conversion up

Where video budgets actually go

Ask a marketing team what video costs and they will name a day rate. Ask them what a finished video cost and the number is usually two to three times higher, with most of the gap sitting in places nobody budgeted for: revision cycles, reshoots caused by a script that was never signed off, and the four weeks of elapsed time between the shoot and anybody actually publishing anything.

A rough but honest breakdown of where money goes on a typical branded piece, once everything is counted:

The pattern is consistent: money is lost in post and on reshoots, but the cause is nearly always insufficient pre-production. The single highest-return change most teams can make is to move budget from the shoot day to the two weeks before it.

Pre-production is the whole job

Our rule: nothing gets scheduled until four documents are signed off. Not agreed verbally — signed off, by name, in writing.

1. The one-line purpose

One sentence: who this is for, what we want them to think or do afterwards, and where they will see it. If the answer includes the word “awareness” without further detail, keep pushing. A video for a paid social pre-roll and a video for a sales follow-up email are different films, and trying to make one serve both produces something that serves neither.

2. The locked script

Every word. Including the bits people improvise. Especially the bits people improvise — write them, then let the talent improvise around a known structure. An unlocked script is the leading cause of reshoots.

3. The shot list

Every setup, with lens, framing and approximate duration. The shot list is what converts a script into a schedule. Without it, a day disappears into indecision between takes.

4. The deliverables matrix

Written before the shoot, not after: every version you will need, with aspect ratio, duration, caption requirement and where it will run. This document changes how you shoot — if you know a nine-by-sixteen cut is required, you frame with headroom for it on the day, which costs nothing, rather than cropping badly in post, which costs plenty.

Writing for the ear, not the page

Marketing copy read aloud sounds like marketing copy read aloud. The rewrite from page to ear is a distinct craft step and it is routinely skipped.

• Short sentences. A speaker needs somewhere to breathe. Any sentence you cannot say in one breath needs cutting.
• One idea per sentence. Subordinate clauses that work fine on a page are lost in audio.
• Concrete nouns. “Optimise your workflow” evaporates. “Cut two days off your month-end close” does not.
• Say the thing in the first eight seconds. Watch-through curves are brutally consistent about this. The steepest drop is always at the start, and it is almost always caused by a preamble.
• Write for muted playback. A large share of social video is watched without sound. If the piece collapses without audio, it needs on-screen text carrying the spine of the argument.
• Read it aloud, standing up. Every awkward construction announces itself within one pass.

Shoot day: fewer setups, more takes

The instinct on a shoot day is to maximise coverage — as many setups, angles and locations as the schedule allows. This produces footage that is broad and shallow, and an edit where nothing quite cuts together.

The alternative: cut the number of setups roughly in half, and spend the recovered time getting each one properly. Three excellent setups edit into something watchable. Nine mediocre ones do not.

Practical habits that consistently save a day:

• Shoot the most important thing first, while everyone is fresh and before the schedule starts slipping. It always slips.
• Roll long on interviews. The usable answer is usually the third one, after the subject has stopped performing.
• Capture five minutes of room tone. Costs nothing on the day, saves hours in post.
• Shoot cutaways deliberately, from the shot list, not as an afterthought in the last twenty minutes.
• Frame for the crop you know is coming. If a vertical cut is in the deliverables matrix, leave the room for it.
• Review one take of every setup on a proper monitor before moving on. Focus problems found on set cost minutes; found in post they cost a reshoot.

The edit rhythm that holds attention

Retention curves have a recognisable shape. There is a cliff in the first five seconds, a shallower decline through the middle, and a second drop wherever the piece changes subject without signalling it.

What consistently flattens the middle section:

1. Open on the most interesting frame you have. Not a logo, not an establishing shot. The thing that makes somebody stop.
2. State the payoff within eight seconds. What will the viewer know or be able to do by the end.
3. Cut on movement. Cuts landing on a gesture or a change in delivery are close to invisible; cuts landing on stillness are felt.
4. Signal transitions verbally. A spoken “so here is the part that surprised us” holds far more attention than a graphic wipe.
5. Burn in captions. Not auto-generated ones — corrected ones. Names and technical terms are always wrong in automatic captions, and always in the most visible places.
6. End on a single, specific action. One. “Subscribe, and check the link, and follow us” produces none of the three.

One shoot, eleven assets

The economics of video only work when a single production day feeds a quarter of publishing. Planned from the start, one interview-led shoot reliably yields:

• The primary film — two to three minutes, site and email
• A thirty-second cut for paid social
• Three to five vertical clips, fifteen to forty seconds, each making one point
• A silent, caption-led version for feed placements
• An audio-only cut for podcast or ad use
• A transcript, edited into a written article
• Six to ten pull-quote stills for static social
• Stills and B-roll for the site and future decks
• An animated data sequence, if the interview contained figures
• A sales-enablement cut with the marketing framing removed
• A behind-the-scenes cut from the same day's phone footage

None of that is extra shooting. It is entirely a matter of deciding in advance, so that the material exists and the edit is set up to produce all of it in one pass rather than eleven.

Which video does which job

Two observations from running this table with a lot of clients. The formats with the lowest costs — product walkthroughs and batched expert pieces — are consistently the ones that influence deals most directly, and consistently the ones teams make last. And the brand film, which absorbs the largest single share of most video budgets, is the hardest of all to attribute anything to.

Measuring video honestly

View counts are close to meaningless on their own, because platforms define a view differently and generously. More useful:

• Retention at the halfway mark. The most honest single indicator of whether the piece works.
• The shape of the drop-off curve. A cliff at a specific timestamp tells you exactly which sentence lost people. Watch it back and fix that sentence in the next one.
• Replays and rewinds. Usually a sign of a section that is unclear rather than compelling. Worth investigating before celebrating.
• Assisted conversion. Did people who watched behave differently afterwards? Almost never a clean number, but the direction is informative.
• Sales usage. The cheapest quality signal available: if the sales team sends a video to prospects unprompted, it works. If they never touch it, it does not, whatever the view count says.

That last one is worth more than most dashboards. Ask the sales team once a quarter which video they actually send. The answer is often not the one that cost the most.