Pre Production: The Stage That Saves the Most Money

What actually happens during pre production, why every hour spent there removes several later, and the documents a project should have before anything is filmed.

Pre production is the stage clients most want to shorten and the stage that most reliably determines what a project costs. The reason is arithmetic rather than philosophy. A decision changed during pre production costs an hour of someone's time. The same decision changed during a shoot costs a crew day. Changed after rendering, it costs a week. The cost of a decision rises by roughly an order of magnitude at each stage boundary, so the stage where decisions are cheapest is the stage worth extending.

The first pre production output is a locked brief, and locked is the operative word. It names the audience, the message, the runtime, the channels, the mandatory legal and brand elements, the deadline and, critically, the person who approves. Projects with an unnamed approver discover their approvers sequentially, each one after work has been done, which is the single most common cause of budget overrun in corporate video.

The second output is the script, and its most useful property is that it fixes the runtime. Voiceover runs at roughly one hundred and forty to one hundred and sixty words per minute for a comfortable commercial read, which means a sixty second film carries about one hundred and fifty words. Clients regularly arrive with four hundred. Resolving that gap in pre production is a conversation. Resolving it in the edit means either a rushed read that sounds anxious or a film that runs long and gets rejected by the platform it was made for.

The third output is the storyboard or shot plan, which converts words into images and durations. Every panel should state what is on screen, how long it holds, what the camera does and what the viewer should understand from it. Storyboarding exposes problems that scripts conceal, most commonly a line of narration with nothing to show against it, and a sequence of ideas that reads logically but has no visual progression. It is also the last document that can be changed for the cost of an afternoon.

The fourth output is the logistics package: schedule, call sheet, location agreements, permits, casting, equipment list, catering, transport and insurance. In Malaysia the permit line deserves particular attention because lead times for public spaces, malls, transport hubs and drone operation are measured in weeks. A schedule that assumes a permit will arrive in three days is a schedule with a fault built into it.

For AI and hybrid projects the equivalent pre production work is asset preparation and visual development: gathering brand files in vector form, CAD where relevant, product references, and running the still frame exploration that fixes the visual language. This is genuinely the same stage doing the same job. It establishes what the finished thing will look like while changing it is still cheap, and it produces the reference set that everything downstream is checked against.

The risk register is the document nobody asks for and everybody needs. It lists what could go wrong, how likely it is, what it would cost, and what the contingency is. Rain on the only outdoor day. A client sample arriving damaged. Talent falling ill. A venue cancelling. An approver being on leave during the review window. Most of these have cheap mitigations if identified in advance and expensive ones if discovered on the day. Mirzaei et al. (2025) make the case that project methodologies should be customised to the specific project rather than applied uniformly, and a risk register is how that customisation gets decided rather than improvised.

Approval sequencing is the piece of planning that most often goes missing. A project should know, before it starts, who reviews the script, who reviews the storyboard, who reviews the offline edit, who signs the final, how long each review has, and what happens if a review is late. Vakratsas and Wang (2021) framed AI as operating inside the creative process rather than replacing the judgement that governs it, and the same applies to schedule: no amount of production speed compensates for an approval chain that takes three weeks to respond.

There is a reasonable question about how much pre production is enough, and the answer scales with irreversibility. A single day social shoot that can be repeated next week needs light planning. A launch film tied to an immovable event date, involving a location that is available once, needs heavy planning. The test is what happens if a decision turns out to be wrong: if it can be redone cheaply, plan lightly; if it cannot be redone at all, plan until the risk is genuinely covered.

For a client, the practical measure of whether pre production was adequate is what the shoot day feels like. A well planned production day is uneventful, slightly boring, and finishes on time. A day full of energy, improvisation and problem solving is usually a day paying for decisions that were not made three weeks earlier.

References

Mirzaei, M., Mabin, V. J., & Zwikael, O. (2025). Customising hybrid project management methodologies. Production Planning & Control, 36(9), 1188–1205. https://doi.org/10.1080/09537287.2024.2349231

Vakratsas, D., & Wang, X. S. (2021). Artificial intelligence in advertising creativity. Journal of Advertising, 50(1), 39–51. https://doi.org/10.1080/00913367.2020.1843090