# How to Write a Documentary Script for Narration-Led Video > Write a documentary script with a clear thesis, evidence, hook, acts, scene intent, spoken rhythm, and a reviewable source trail for AI-assisted production. Updated: 2026-07-11 Audience: YouTube documentary writers, researchers, producers, and faceless creators Canonical: https://onira.studio/guides/how-to-write-a-documentary-script ## Key takeaways - Write one thesis that every act advances or complicates. - Attach evidence and visual intent to the outline before drafting prose. - Write for breath, emphasis, and the viewer's changing understanding. ## Begin with a question and a defensible thesis The opening question creates curiosity; the thesis gives the film direction. 'What happened at Pompeii?' invites a summary. 'What did ordinary routines reveal about the final hours before Vesuvius?' creates a human-scale investigation with a specific visual world. Write the thesis in one or two sentences and list the evidence needed to support it. Mark uncertainty and disagreement instead of smoothing them into false certainty. A documentary earns trust by showing the limits of what is known. - Question the viewer wants answered. - Claim the film can actually support. - Uncertainty the narration must preserve. ## Outline acts as changes in understanding An act should change what the viewer believes, not merely move to another subtopic. The opening establishes the familiar picture and the central tension. The middle introduces evidence, conflict, or a hidden mechanism. The final act resolves the question while acknowledging what remains uncertain. For each beat, record the claim, source, emotional function, and possible visual. This prevents the finished script from becoming a voice essay with generic imagery added afterward. - Hook: a concrete contradiction, image, or unanswered question. - Development: evidence that deepens or overturns the initial assumption. - Resolution: consequence, meaning, and a precise final image. ## Write for the ear Spoken language needs more orientation than written prose. Use concrete nouns, active verbs, and sentences that can be understood once. Vary sentence length, but avoid stacking dates, names, and subordinate clauses in one breath. Read every paragraph aloud. Mark pauses, pronunciation, and words that need emphasis. If a line cannot be spoken naturally, it will not become clear simply because an AI voice reads it smoothly. - One principal idea per sentence. - Transitions that tell the listener where the argument is moving. - Specific sensory detail that can become a shot. ## Separate narration from visual direction Narration explains, argues, and guides attention. Visual direction shows evidence, place, action, scale, texture, or contrast. Do not make the voice redundantly describe everything on screen, and do not ask generated visuals to prove a claim they cannot reliably depict. Create a two-column review: what the viewer hears and what the viewer needs to see. Flag maps, diagrams, readable documents, exact interfaces, and archival evidence for a different production method when generative imagery would be misleading. - Use generated reconstruction as illustration, not documentary evidence. - Reserve exact factual visuals for verified assets. - Give each scene one dominant visual purpose. ## Publication checklist - The central question and thesis are explicit. - Every major claim has a source or is labeled as interpretation. - Each act changes the viewer's understanding. - The script has been read aloud and pronunciation checked. - Narration and visual direction are reviewed separately. - Reconstruction is never presented as authentic evidence. ## Sources - [US National Archives primary-source guidance](https://www.archives.gov/education/research/primary-sources) - [Library of Congress primary-source guide](https://www.loc.gov/programs/teachers/getting-started-with-primary-sources/) - [Google guidance on helpful, reliable content](https://developers.google.com/search/docs/fundamentals/creating-helpful-content?hl=en) ## Questions ### How long should a YouTube documentary script be? Write to the target spoken duration and then measure the actual narration. Speaking rate varies by narrator, language, subject, and intended pace, so a word-count formula is only a rough planning aid. ### Can AI write a documentary script from one prompt? AI can draft a structure and prose, but the creator still needs to define the thesis, constrain the sources, check claims, preserve uncertainty, and revise for spoken rhythm and visual feasibility. ### Should the script include shot descriptions? Maintain linked visual intent for each beat, but keep final image and motion direction in dedicated production fields. This makes it easier to revise the spoken argument without confusing it with generation prompts. ## Product boundary - Onira delivers a final MP4; it does not upload or schedule posts on YouTube or social platforms. - Onira provides a reviewable production workflow; creators remain responsible for approving the story, facts, rights, disclosure, and final publication. - Director chat is limited to regenerating one selected PREVIEW timeline video clip; other available Studio controls are separate direct actions. - Creators must review facts, sources, rights, realistic-synthetic-media disclosure, and platform policy before publishing. - Onira does not guarantee YouTube monetization, reach, factual accuracy, or legal clearance.