Choosing an Advertising Film Structure Starts with the Same Question

Advertising Film Strategy Research 02/10 · ONCE Editorial

In an advertising proposal meeting, “a brand story has more stature,” “a product demonstration is more direct,” and “vertical video suits younger people” can appear in the same conversation. The team seems to be comparing film types, but may actually be comparing communication objectives, expressive structures, media formats and production techniques. As the discussion grows livelier, the audience judgment it should change can become less clear.

This article asks which conditions make another expressive structure worth testing against the familiar choice. ONCE's central argument is to make two proposals carry the same promise, then compare how each helps an audience understand, believe or retrieve that promise. Treat a film structure as a candidate mechanism rather than an industry identity. Only when the comparison and outcome are clear does “a story film suits us better” become a planning judgment that can be challenged.

The previous article, testing the marketing bottleneck before production, examined whether communication deserves to address the problem. This article starts from an established communication task and considers expression. It does not assume every project needs another complete marketing diagnosis.

Separate four classification axes

“Should we make a TV commercial or use 3D?” is an incomplete choice. A TV commercial usually describes an advertising placement or project format; 3D describes a production method. One commercial can combine a character story, product demonstration and 3D explanation. Portrait or landscape framing and running time shape visual and information capacity, but do not independently define a persuasive structure.

A planning conversation should separate at least four decisions. The communication task defines whose judgment should change and how. The expressive structure organizes information through events, processes, arguments or statements by people. The media version establishes the entry point and opportunity to watch. Production techniques determine how the images will be realized. Brand films, product films and corporate films can help define project scope, but are also not mutually exclusive psychological mechanisms.

For example, “help people new to backup power judge whether it fits their actual outage situation” is a communication task. Showing the connection process first or starting with a family's choices during an outage offers two candidate structures. Making either horizontal or vertical is a separate decision. Live action, 3D and generative AI should serve the facts that need accurate representation and the conditions under which production is feasible. Technical novelty does not establish that viewers have made the intended judgment.

This does not dismiss industry experience. Experience can narrow candidates and reveal filming difficulties. It cannot replace the question of whether the conditions behind earlier success still apply. In one industry, launch awareness, installation comparison and support explanation may need different structures. One structure can also serve quite different marketing tasks.

Recent research does not give us a ranking of film types

Shehu and colleagues' 2026 study analyzes 3,495 television, online and social video ads, coding 161 features with humans and algorithms and using Bayesian Additive Regression Trees to examine recall and attitudes. The official abstract reports differences across channels and objectives. We accessed only the abstract and publication information, not the complete market, period or uncertainty details. We therefore reproduce no feature rankings or causal sales claim.[1]

Raithel and colleagues' 2026 study matches 280 US Super Bowl brand-year events from 2013–2022 to 17,080 daily observations, modeling pre/post-advertising changes. Differentiating messages are associated with improvements in earlier brand metrics; purchase measures are survey reports. Identification depends on trend and exogeneity assumptions, with prereleases and unobserved marketing acknowledged as possible confounds. It is not random assignment to stories or demonstrations.[2]

The materials help separate questions. The first examines content-feature relationships across channel and objective; the second examines whether advertising carries information that differentiates the brand. Neither randomly compares “emotional,” “specification” and “testimonial” films for one brand under equivalent media opportunities. Using them to endorse a universal ranking would exceed their scope.

ONCE's bounded interpretation is to check whether a structure provides the information required for the intended judgment, alongside whether people like the content. Liking may be a process condition. If the task is recognizing compatibility limits, the primary result should be correct judgment. If it is brand attribution, check whether viewers assign the content to the right brand. Do not switch to a flattering metric after seeing results and announce a winner.

An experiment in one category does not make one expression best everywhere

Papadopoulou and colleagues' paper appeared in Psychology & Marketing in 2025, first published November 26, 2024. We read relevant UK menstrual-care static-material methods and results. Study 2b randomly allocated 624 women across informational/transformational messages and concrete “how”/abstract “why” combinations, finding a purchase-intention interaction. This was not film or sales evidence. Study 2a's fictitious-brand choice involved a lottery whose prize became cash, not an actual order.[3]

The experiment warrants considering whether expression fits the thinking task at that moment. It does not warrant “women's products require emotional films.” Transfer from static materials to film requires viewers to hear or see the expression clearly, have an opportunity to process key information, and encounter an execution that preserves the intended comparison. Country, category and situation cannot simply disappear.

A further competing explanation is that a story version wins because its promise is stronger, wording clearer or performer more credible, rather than because of the “story” label. An experiment can compare specific materials. Planning must state whether it aims to generalize a result for those executions or a more abstract structure. Those conclusions require different evidence.

Give the same uncertainty competing proposals

The following is ONCE's comparison tool, not an empirically validated effectiveness ranking. Write the audience's uncertainty first, then choose two structures that could address it. That is usually more useful than producing all seven types.

The final row is a business content task within which the preceding structures can be combined. Interviews can sit inside stories; demonstrations can evoke emotion. Assess combinations by the judgment each segment adds, rather than the number of types included.

Suppose a backup-power brand wants first-time buyers to judge whether a device fits their actual household loads. This is a hypothetical planning situation, not an ONCE client experiment. Proposal A starts with a household outage and brings the device into the action. Proposal B starts with compatibility conditions and the connection process before returning to the household outcome. Both must use verified facts, the same limits and the same next-step entry point.

If the main obstacle is not recognizing a need, A merits consideration. If viewers already have a need and are checking compatibility, B merits consideration. These remain predictions to test. A situation-based story can explain limits clearly, and a demonstration can activate a need. The actual differences concern which information gets attention earlier, what depends on preceding material, and who can complete that judgment.

Prior knowledge, involvement and risk therefore cannot become age labels. An unfamiliar, consequential choice may require concrete conditions, yet a high price does not mean people can process every specification in a feed. A low-risk purchase may still require demonstration because the use is unclear. Specify the choice people are making now, what they already know and what concerns them, then decide information density and order.

Use a public work to understand combination without inferring effects

The public case text for Huawei Mall's Mother's Day commercial connects a mother's everyday “choosing” with her family's online order and gift of a phone. We read the Chinese and English case pages only, without completing video or sound analysis. We obtained no client brief, exposure, sales or experiment records.

This limited observation illustrates that story and functional organization can coexist: ordering becomes the family's act of choosing for the mother. ONCE's interpretation is that this deserves consideration where the task concerns the meaning of a seasonal gift. Comparing particular devices' interfaces or performance would require a different information arrangement. We do not infer which actual bottleneck the case addressed or treat it as proof that stories outperform demonstrations.

ONCE's TV commercial and campaign production approach provides a next step for checking how the proposition, key action and versions enter production scope. It is a project-selection resource, not research evidence. Once a structure enters a proposal, confirm whether actual product actions, performers' tasks, settings, interface assets and post-production can carry the promise. Creative validity and production feasibility are connected but separate judgments.

Decide what the comparison is meant to estimate

If two complete ads are randomly shown to different viewers, the difference can come from the entire creative package: structure, cast, music, wording and finish. This can help decide which film to release. It does not automatically answer whether storytelling is better. Conversely, mechanically constructed samples that isolate one variable may lose the persuasive qualities of real creative work.

We therefore suggest separating two levels. The project decision asks which feasible proposal better completes the same task under specified viewing conditions. Learning about the mechanism asks which information sequence or argument arrangement creates a change. Do not use the first result to answer the second automatically. Nor must every commercial decision stop until a perfectly isolated mechanism can be established.

The following is an experiment proposal that has not been implemented. For the hypothetical backup-power project, recruit people with an actual need, record prior knowledge, and randomly allocate comparably finished rough cuts. Hold the brand promise, true facts, price, running-time range and subsequent entry point constant. Offer the same viewing opportunity and record exits. Do not compare only people who finish: finishing may itself result from the expression.

Prespecify correct judgment of one compatibility condition as the primary outcome, while also observing brand attribution and understanding of relevant benefits. Use open responses to see what viewers learned, not just whether the film was good. If observing requests for a suitability check, provide the same genuine process and distinguish the request from purchase. Incorrect understanding of a critical limit should count as substantive failure rather than be offset by greater liking.

If A is more memorable but produces more compatibility errors, retain the character-led entry while rebuilding the conditions explanation. If B improves understanding but many people exit, consider reducing its entry burden or changing the media context. These are proposed update rules, not results. Small samples, recruitment bias, unequal finish or wide uncertainty intervals require uncertainty to remain visible. Repeatedly testing until one version becomes “significant” can also amplify mistaken judgments.

With resources, a later round could change information order within one structure and check more than one execution. That helps distinguish a performer's or a sentence's isolated success. Low-cost rough cuts can reveal comprehension and expression problems, but cannot directly predict incremental sales under real media exposure. Estimating purchase changes requires actual exposure, a credible comparison, sufficient observation and transaction records. This article does not claim ONCE has such a system or experimental results.

Put the selection rationale into the production brief

When approving the choice, a team should be able to state the specific uncertainty it faces, how the structure attempts to change judgment, the facts viewers must see clearly, and the result that would prompt a different choice. Translate that into production. An important operation needs more than decorative close-ups; clothing alone cannot establish a source's expertise; a story's turn cannot depend on editing away limitations.

Under a tight budget, protect the presentation on which the promise depends before evaluating the added value of visual spectacle. Conversely, if characters, setting and sound are needed to experience the difference, not every production investment is removable packaging. There is no universal answer that cheaper work feels truer or cinematic scale works better. Ask what the cut component actually contributes.

Back in the proposal meeting, the useful disagreement concerns which audience judgment is worth changing and why the two arrangements might yield different results. It need not end at a preference for emotion or specifications. Industry practice supplies candidates, research supplies constraints, and comparison supports learning. ONCE's limited principle is to give one task two expressive proposals that could be overturned before choosing which deserves production.

For a project, send ONCE the communication task and two candidate expressions, together with audience uncertainties, verified product facts and planned entry points. These materials help turn the selection rationale into a script and production scope.

Materials and references

Searches and link checks are current to October 5, 2026, prioritizing original materials from the previous 24 months. This is research-based analysis, not a peer-reviewed paper, systematic review or original experiment. We read [1]'s official abstract and publication information, relevant methods, measures, results and limitations in [2], relevant experimental methods and results in [3], and the ONCE page text discussed above. We did not test mainland-China network access. The Chinese article retains overseas titles and DOI text; this version links to original material successfully read during this review. No retraction notice appeared on the accessed pages; we did not conduct a comprehensive registry search.

[1] Edlira Shehu, Prasad A. Naik, Ziad Elmously, Rick Candelaria. Creative synergy: leveraging AI and human coding to enhance recall and attitudes in TV, social and online video advertising. Journal of Marketing Analytics, online July 4, 2026. DOI: 10.1057/s41270-026-00498-8. University publication record and abstract accessible; complete text is subscription content and was not accessed.

[2] Sascha Raithel, Alexander Mafael, Bastian Amberg, Charles R. Taylor, David W. Stewart. The missing big idea: Brand-differentiating messages and advertising effectiveness across the brand funnel. Journal of the Academy of Marketing Science, July 7, 2026. DOI: 10.1007/s11747-026-01166-0. Original empirical research with open full text; not a randomized film-structure comparison.

[3] Christina Papadopoulou, Magnus Hultman, Pejvak Oghazi. Authenticity Perceptions of Informational and Transformational Advertising: Decoding the Role of Construal Level Mindset. Psychology & Marketing, 42(3):817–841, 2025; first published November 26, 2024. DOI: 10.1002/mar.22153. Relevant open-text sections were read through the web reader. A direct link is omitted because the final ordinary request returned 403 and the bounded DOI fallback was unavailable. Static materials and intention/choice tasks are not a film-advertising order experiment.