A Quality Control Checklist Before Any AI Video Is Delivered

The specific checks that catch the defects generative production reliably produces, ordered so that the cheapest fixes are found first.

Generative production has a predictable set of defects, which means quality control can be a checklist rather than a general inspection. Running the same checks in the same order on every project catches most problems before a client sees them, and catches them at the stage where they are cheapest to fix.

Start with the brand critical elements, because these are the most damaging and the most fixable. Every logo, wordmark, product name and piece of packaging copy in the film should be checked at full size, frame by frame where it appears. Anything generated rather than composited is likely to be subtly wrong. Kirk and Givi (2025) found that perceptions of authenticity shape consumer responses to AI generated marketing communications, and a nearly correct brand mark is exactly the detail that converts a viewer from impressed to suspicious.

Next check the product itself for accuracy: proportions, colour, materials, finish, and configuration. Compare against a physical sample or the specification document rather than against memory. This matters commercially rather than aesthetically, because a customer who receives something different from what the film showed has a concrete complaint.

Then look for temporal drift across every shot. Play each shot at full speed and watch whether the subject's proportions, the colour or the background change from the first frame to the last. Drift is the characteristic generative defect and the fix is almost always to shorten the shot to its stable portion rather than to attempt a repair.

Check hands, faces and articulated motion specifically, because these fail in ways that are obvious to an audience and easy to miss in a review focused on the overall film. Pause on any frame where a hand is visible. If it is wrong, the options are to reframe, cut earlier, shorten, or regenerate, and the decision should be made rather than deferred.

Check physical plausibility across each frame: do shadows fall in consistent directions, do reflections correspond to something in the scene, do objects intersect correctly, does anything float. Audiences frequently cannot name these errors and reliably feel them. Errors in secondary elements can be masked or darkened; errors in the subject usually require the shot to be replaced.

Then check consistency across the whole sequence rather than shot by shot. Play the film through and watch for changes in light direction, colour temperature, lens character, grain and sharpness between adjacent shots. Uneven enhancement is a common cause: if some shots were upscaled or denoised and others were not, they will differ in micro texture regardless of grading.

Check the text layer against a source document. Every name, title, figure, claim, legal line and call to action should be verified against something authoritative rather than against the script, which may itself contain an error carried forward. This five minute check prevents the most embarrassing category of delivery error.

Check the audio properly rather than assuming it is fine because it sounded right during the edit. Narration intelligibility over the music, no clipping, consistent levels across the film, and loudness appropriate to the destination. Zhang et al. (2025) found that audiovisual features of short video advertising contribute measurably to consumer engagement behaviours, and Xiao et al. (2026) reached compatible conclusions from a combined visual and audio perspective, which is why this is a delivery check rather than a refinement.

Finally, check each deliverable in its actual destination rather than in a player. The social cut on a phone with the sound off in daylight. The event version on a large screen at show brightness. The web version in a browser. Each of these reveals a different class of problem, and none of them is visible on a calibrated monitor in a dark room, which is where every other check was performed.

References

Kirk, C. P., & Givi, J. (2025). The AI-authorship effect: Understanding authenticity, moral disgust, and consumer responses to AI-generated marketing communications. Journal of Business Research, 186, Article 114984. https://doi.org/10.1016/j.jbusres.2024.114984

Zhang, Z., Qiu, K., & Ye, Y. (2025). Influence of audiovisual features of short video advertising on consumer engagement behaviors: Evidence from TikTok. Journal of Business Research, 201, Article 115662. https://doi.org/10.1016/j.jbusres.2025.115662

Xiao, L., Li, X., & Mou, J. (2026). Exploring user engagement behavior with short-form video advertising on short-form video platforms: A visual-audio perspective. Internet Research, 36(1), 154–188. https://doi.org/10.1108/INTR-07-2023-0521