AI video advertising in Morocco: a practical production guide

A production brief connects commercial intent to material that can actually be made. Record the product or service, mandatory facts, approved spelling, offer conditions, brand tone, call to action, required languages, and elements that must not appear. Separate verified claims from creative language. If a promise depends on availability, geography, eligibility, or a limited period, place that condition beside the script instead of adding it at the final export.

Define the desired duration as a range or placement need, not as a substitute for the story. Include reference work for direction, but describe what is useful about each reference: pace, framing, texture, product visibility, or narration. A reference is not permission to copy another brand, performer, character, soundtrack, or protected visual identity.

• One primary message, one supporting proof point, and one clear call to action.
• Required product details, on-screen copy, pronunciation, and factual review owner.
• Visual boundaries, cultural considerations, disclosure needs, and brand-safety exclusions.
• Delivery date, approval dates, planned review rounds, and final publishing owner.

Create a source asset inventory covering product images, footage, logos, fonts, packaging, locations, people, voice references, music, existing campaign files, and past brand material. For every file, record its source, resolution, owner, intended use, and whether approval is documented. This prevents a low-quality reference or uncertain asset from quietly becoming the foundation of the finished advertisement.

The inventory should distinguish supplied files from newly created material and generated output. A tool producing an image, voice, or scene does not by itself prove that commercial usage rights are available. The commissioning team should confirm permissions and, when needed, obtain qualified advice. Identifiable people may also involve privacy and personal-data obligations; in Morocco, the CNDP materials are a useful primary starting point rather than a substitute for advice about a specific campaign.

• Likeness and image rights: appearance, identifiable faces, performers, customers, and reference people.
• Voice and audio: narration, dubbing, voice clone references, music, sound effects, and recording licences.
• Brand assets: logo, typography, packaging, trademarks, partner marks, and approved brand guidelines.
• Consent and usage rights: territory, channels, duration, editing permission, and evidence of authorization.

Move from the brief to a script, storyboard, and shot list before producing many variations. The script defines the spoken and written message. The storyboard defines sequence, framing, product presence, and the relationship between generated scenes and supplied material. The shot list turns those choices into production units that can be accepted, revised, or replaced without reopening the entire concept.

Give factual claims a separate review. Product features, prices shown inside supplied packaging, comparisons, testimonials, health language, environmental statements, and offer conditions should be checked by an accountable person. A generated frame may look convincing while adding a label, object, location, or product behavior that is not true. Human review should compare each important frame with the approved brief and source assets.

• Concept approval confirms the central idea and its fit with the campaign objective.
• Script approval confirms copy, pronunciation, claims, conditions, and call to action.
• Storyboard approval confirms scene order, visual references, disclosures, and product representation.
• Rough-cut approval confirms pacing, captions, audio, continuity, and required corrections.

Plan a format family rather than cropping one master at the end. Common working canvases include vertical 9:16, square 1:1, and landscape 16:9, but the correct aspect ratio, duration, file size, safe zone, and text treatment depend on the selected placement. Platform specifications change, so the team should re-check the current official requirements when the media plan is confirmed and again before final delivery.

Important information must remain legible when interfaces cover part of a frame or when a placement starts without sound. Prepare captions or subtitles, review contrast and reading time, and decide whether a transcript is needed. Accessibility is also a production control: it exposes unclear narration, crowded copy, and meaning that depends only on audio. French, Arabic, and other language versions need human language review rather than automatic text replacement alone.

• Confirm intended placements and current specifications with the publishing or media owner.
• Protect faces, product details, disclosures, and the call to action inside practical safe zones.
• Review captions, subtitles, transcript, contrast, pronunciation, and right-to-left layout where applicable.
• Export and name each master by language, ratio, duration, version, and approval status.

AI-assisted production can create many plausible alternatives quickly, which makes approval discipline more important. Name one person who consolidates feedback and one person who gives final approval. Scattered comments from several channels produce contradictory edits and make it difficult to know which version is current. A short decision log should record what changed, why it changed, and who accepted it.

Agree on review rounds and what each round is for. Early iterations should resolve the concept, script, and visual direction; later versions should correct execution rather than introduce a different campaign. If the brief changes after approval, treat the impact on schedule, shots, formats, rights checks, and voice work as a scope decision. Keep rejected generated scenes out of the delivery folder so they cannot be published accidentally.

• Concept, script, storyboard, rough cut, language review, and final masters have named approval gates.
• Version names identify date, language, aspect ratio, review status, and change history.
• Feedback is consolidated, attributable, and tied to the approved brief.
• A final human review checks every export rather than assuming one approved master covers all adaptations.

A responsible estimate is built from scope, not from a universal per-video price. Major cost drivers include concept development, script complexity, the number and difficulty of shots, source-asset readiness, cleanup or compositing, narration and music, languages, captions, aspect ratios, durations, disclosure work, review rounds, and the number of approved versions. Media buying, talent agreements, stock licences, and landing-page work may be separate items.

Ask the proposal to state assumptions and exclusions. A small source library may require more asset creation; a regulated claim may require more review; several placements may require distinct edits rather than mechanical crops. Clarify which iterations are included, who pays for third-party licences, what triggers a change request, and which files are handed over. This lets competing proposals be compared on equivalent responsibilities.

• Creative scope: concepts, scripts, storyboards, shots, durations, and localized variants.
• Production inputs: supplied assets, new generation, editing, compositing, voice, music, and captions.
• Governance: approvers, review rounds, factual checks, disclosure checks, and version management.
• Delivery: masters, source files, usage scope, licences, archive period, and later adaptations.

Measurement should follow the campaign objective. Decide before launch which event matters, how it is tracked, and which system is the source of truth. Creative tests can compare hooks, openings, messages, calls to action, or lengths, but change a limited set of variables so the result can be interpreted. Keep campaign naming and version naming aligned so reporting can identify the asset that actually ran.

No video can guarantee ROAS, conversion, or another business outcome. Results also depend on media buying, audience selection, auction conditions, offer strength, landing-page experience, sales follow-up, seasonality, and measurement quality. Treat early performance as evidence for the next test, not proof that one isolated creative decision caused the entire result. When attribution is incomplete, report that limitation rather than presenting estimated outcomes as observed facts.

• Define the primary metric, supporting diagnostics, tracking owner, and decision window.
• Map each published creative ID to its language, format, hook, message, and landing destination.
• Record media, targeting, offer, and landing-page changes that could affect interpretation.
• Decide in advance whether the next action is keep, revise, stop, or run a clearer test.

Before publishing, review whether generated or altered content needs disclosure or labeling under the current platform, campaign, contractual, or regulatory context. Do not present a synthetic person as a real customer, employee, expert, or endorser. A UGC-style treatment should be described accurately and must not turn a capability demonstration into fabricated testimony. Review political, medical, financial, children-related, or other sensitive contexts with additional care.

Brand safety also includes what the model adds unintentionally: distorted logos, unsafe product use, misleading settings, cultural mismatches, implausible text, or a resemblance to a recognizable person. Check every final frame, voice track, caption, thumbnail, and localized export. Platform policies and disclosure tools evolve, so production teams should verify current official guidance at release time rather than relying on an old checklist.

Close the project with a written handoff. List approved masters, language and aspect-ratio variants, captions, transcript, thumbnails, source files included, font and music information, licence records supplied, and the final script. Record the usage scope, channels, territories, usage period, and any limitation attached to talent, voice, music, stock, or partner assets. The client should know which account stores the files and who controls future access.

The handoff should also identify generated content, the approved version, and material that must not be reused. Keep a compact decision and change history so a later adaptation does not revive a rejected claim or expired asset. If editable project files are excluded, say so before production. A clear archive and passation process makes future resizing, localization, and campaign learning safer and less dependent on individual memory.