Advertising Film Strategy Research 01/10 · ONCE Editorial
“The product is not selling well enough. We need a more compelling film.” That sounds reasonable, but it combines two untested judgments: communication is the main cause of the sales problem, and a more compelling execution can change it. Once the conversation moves to formats, casting and visual style, those judgments can disappear from view.
This article asks which marketing bottleneck a team should investigate before committing a production budget. ONCE's central argument is that a production task should rest on a falsifiable intervention hypothesis: for a particular audience in a particular situation, what change in information, judgment or conditions for action could change the outcome? A sales objective tells us where we want to go. It does not tell us what the camera should show.
Separate the three tasks first
A business objective might be securing product trials, generating qualified enquiries, increasing repeat purchases or entering a new procurement shortlist. A communication task specifies the change required in the audience: understanding installation, believing a verified performance claim, or retrieving the brand when a particular need arises. Production deliverables are vehicles for that change: a hero film, demonstration, interview, versions and footage. These tasks are connected, but they are not interchangeable.
Imagine a cleaning-equipment company seeking more trials. This is a hypothetical example for comparing explanations, not an ONCE client study. If prospective buyers cannot tell whether the equipment fits their space, showing its dimensions and installation may deserve testing. If they understand it but worry about upkeep, the script should consider the actual maintenance process and evidence for its cost. If they know it fits and want to try it but their area is not served, additional emotional scenes cannot create service coverage.
The same symptom—too few trials—leads to three different production tasks, or even a decision to delay production. Planning creates value by reducing that misallocation, rather than attaching a marketing term to a film already approved.
A usable task statement could be: “Among renters considering this equipment for the first time who cannot judge whether installation is difficult, test whether an accurate installation demonstration improves understanding and willingness to request a suitability check.” It identifies the audience, situation, judgment to change and observable outcome. Whether it ultimately generates additional orders requires a further level of evaluation.
Ask about conditions for behavior before applying a funnel
Michie, van Stralen and West's 2011 behaviour change wheel paper describes the COM-B system of capability, opportunity and motivation. It developed a framework through a review and classification tests, rather than an experiment showing that films increase purchases. We read the relevant methods and definitions and use it to generate questions, not as a validated advertising diagnostic model.[1]
Whether people know how to assess an offer and act can be examined as a capability question. Whether availability, channels and timing permit action can be examined as opportunity. Need, perceived value and concerns can be examined as motivation. These conditions may interact or coexist; they do not require everyone to advance through awareness, liking and purchase in a fixed order.
A procurement professional might discover a brand while completing a tender task, then check specifications. A consumer might already like a brand but care about installation only when replacing equipment. Calling both “awareness-stage users” does not automatically produce a useful filming decision. This is ONCE's planning interpretation, to be tested in a specific buying situation.
Five explanations and evidence that could challenge each
The comparison below is neither five mandatory stages nor a scoring model. Its purpose is to replace “we should make another film” with explanations that can compete.
| Candidate bottleneck | Discriminating evidence to seek before production |
|---|---|
| Unawareness or failure to retrieve the brand | Can people with a relevant need name options without prompts? After exposure, can they attribute the content to the correct brand rather than merely remember its story? |
| Product misunderstanding | Can people explain the key use, limitations and steps in their own words? Where does misunderstanding occur? |
| Lack of personal relevance | Can people identify a real use situation and its benefit? Separate having no need from failing to see the connection. |
| Distrust or fear of making a mistake | Is the concern about performance, the source, compatibility or support? Does verifiable evidence change that judgment? |
| Willingness blocked by conditions for action | Is the product stocked, purchasable, bookable and deliverable? Are price and procedures within the person's actual acceptable conditions? |
These are research questions, not existing client data. One interview comment—“too expensive”—does not complete the diagnosis. It may reflect a genuine budget constraint, a benefit judged insufficient, or a polite way to end a conversation. The quoted price, alternatives, frequency of need and acceptable conditions must be examined within the same decision.
If people incorrectly believe a product works with every device, strengthening confidence could amplify misinformation. The proper task is to make compatibility limits clear. If some people have no relevant need, even a clear demonstration may not make them an audience worth pursuing. A film can make existing value easier to recognize; expressive power cannot guarantee that everyone receives the same value.
People likely to buy may not be those most open to change
Blake, Nosko and Tadelis studied US eBay paid search in 2012 using advertising suspensions, geographic controls and difference-in-differences; the non-brand experiment also matched historical trends. Frequent buyers responded differently from new and infrequent users, while attributed outcomes differed from incremental outcomes. We read the relevant methods and results in the public 2014 author draft; the corresponding paper appeared in Econometrica in 2015.[2]
This is search-advertising evidence. It does not establish that films for new brands are ineffective or that a particular film structure is best. The narrower implication here is to distinguish people who are easy to reach or likely to click from people whose judgment might change through this communication. Applying that distinction to films still requires testing exposure opportunities, prior knowledge and the content intervention.
Showing a rough cut only to existing followers may produce rapid understanding because they already know the brand, or liking because they support it. That feedback can reveal expression problems, but it does not establish that a new audience's bottleneck has been addressed. Conversely, an unfamiliar viewer's dislike could reflect an irrelevant category rather than a failed story structure.
Before interpreting feedback, record whether participants have a real need, prior product experience and previous information exposure. Define recruitment groups before seeing results, rather than renaming a successful group afterward. If groups are too small or measurements too noisy, acknowledge the uncertainty instead of turning an attractive percentage into a planning conclusion.
An unobserved result differs from an ineffective intervention
Liang and colleagues' 2025 Marketing Science paper reports a large randomized field experiment for an automotive brand on ByteDance. Its official abstract discusses spillover and assessment but does not specify the advertising market, sample or data period. We did not obtain the complete methods, so report no multipliers or generalization to other categories or organic word of mouth.[3]
This material keeps a competing explanation in view: limited platform records could mean an effect occurs through another entry point, rather than that the film has no effect at all. But the possibility of spillover cannot excuse every film. Without evidence outside the platform, unobserved sales cannot simply be credited to advertising.
Planning should specify the intended observation. Searches, saves, enquiries, suitability checks, purchases and repeat purchases are different outcomes. A website click does not establish product understanding; purchase intention does not establish an actual purchase. If the business depends on distributors or lengthy procurement, identify actions that cannot be recorded completely. Those boundaries determine what conclusions are possible and which question to test first.
A published work illustrates expression, not its underlying diagnosis
ONCE's public KTC portable-screen film case describes a story connecting a long-distance relationship with a May 20 declaration of love, using the relationship between the screen and the characters. This article examines the case-page text only. We did not complete a viewing or sound analysis of the film and have no client experiment or sales data.
That limited observation supports a planning proposition: if the task is to explain what a screen means within a relationship, the relationship between people and product could carry that explanation. If the unresolved question concerns interface compatibility or how to move the device, the same narrative may offer insufficient evidence. This is an analysis of a planning choice, not a judgment of the case's actual effectiveness or a claim to know its original client brief.
Once the communication task is clear, ONCE's brand-film and reusable-content planning provides a relevant next step for discussing characters, product actions, sound and versions. The service page informs project selection; it is not academic evidence of any format's sales effect. Expression should follow the task. “Premium brands need emotional films” cannot substitute for that judgment.
A pre-production test: change information and transaction conditions separately
The following is a proposed experiment, not an implemented test or an ONCE result. Using the hypothetical equipment example, first recruit people with a real need who have not purchased and are eligible for service. Record prior knowledge and the situation creating their need, then determine whether the principal uncertainty concerns installation understanding or access to a trial.
Where feasible, randomly compare a standard product introduction with an accurate installation demonstration. Keep price, brand, service area and the next-step procedure the same. Observe primarily whether viewers understand installation requirements, recognize unsuitable conditions and will enter the same suitability-check process. Satisfaction or liking can help interpret responses, but should not replace the primary outcomes.
A separate question is whether offering a genuine, fulfillable trial changes action. With sufficient participants and resources, content and trial conditions could be allocated separately in a crossed design, allowing the team to examine whether they complement each other. With resources for only a simple comparison, answer one question at a time. Do not change the story, price and booking process together and credit every change to the film.
Write the interpretation rules beforehand. If a demonstration improves understanding but not the next action, understanding may not be the main current bottleneck, action costs may remain high, or the test may be unable to detect the difference. If real trial conditions change action while new content shows no clear difference, reconsider the case for more elaborate production. Failure to detect a difference does not prove that no effect exists.
Testing a rough cut or written explanation costs less, but its viewing conditions, finish and media exposure differ from a completed advertising film. It cannot directly predict sales effectiveness. Recruitment, follow-up duration and sharing between people impose limits. Estimating additional purchases still requires real exposure with a credible control or another suitable design, order records and sufficient observations. One small test cannot carry every burden of proof.
Turn the diagnosis into production constraints
After diagnosis, keep at least four items in the brief: the situation in which the audience decides; the judgment that evidence indicates needs changing; the genuine material the film will use to change it; and the result that would lead the team to reject that explanation.
If understanding is the issue, the budget may be better directed toward clear steps, compatibility limits and legible images. If credibility is the issue, prioritize verifiable evidence, appropriate sources and demonstration conditions. If personal relevance is the issue, check the situation and benefit before choosing a story. If the main obstacle lies in supply or transaction procedures, repair those conditions first or communicate them accurately, rather than promise that a film will independently solve them.
A film can contribute to a purchase decision. That does not establish that it addresses the part most worth changing now. ONCE's bounded planning principle is to establish why a communication intervention deserves testing before selecting the expression that will carry it. The more complex the project, the more explicitly that judgment belongs before production, rather than after it in a retrospective.
When preparing a project, send ONCE the product, audience situation and current marketing question. Bringing known facts alongside competing explanations is more useful for defining production scope than supplying only a reference film you like.
Materials and references
Searches and page checks are current to October 5, 2026. We prioritized relevant original research from the previous 24 months and checked necessary foundational materials. This is a research-based analysis, not a peer-reviewed paper, systematic review or original experiment. The reading covered relevant COM-B methods and definitions, relevant experimental methods and results in the eBay author manuscript, the automotive experiment's official abstract, and the ONCE page text discussed above. We did not test access from mainland China. The Chinese article retains bibliographic information and DOI text for overseas sources; this English version links to original materials successfully read during this review.
[1] Susan Michie, Maartje M. van Stralen, Robert West. The behaviour change wheel: A new method for characterising and designing behaviour change interventions. Implementation Science, 6:42, April 23, 2011. DOI: 10.1186/1748-5908-6-42. Open full text; a framework study, not an advertising purchase-effect experiment.
[2] Thomas Blake, Chris Nosko, Steven Tadelis. Consumer Heterogeneity and Paid Search Effectiveness: A Large-Scale Field Experiment. Econometrica, 83(1):155–174, 2015. DOI: 10.3982/ECTA12423. We read the public author manuscript dated April 8, 2014. Its US eBay data concern 2012 paid search and do not transfer directly to film advertising.
[3] Yitian (Sky) Liang, Xinlei (Jack) Chen, Shengnan Han, Jinglong Zhang, Yubo Chen. Is the Money Spent on Short-Form Video Social Platforms Worth It? The Role of Advertising Spillover in a Large-Scale Randomized Field Experiment on ByteDance. Marketing Science, 44(5):1125–1144; online first May 8, 2025. DOI: 10.1287/mksc.2023.0575. Access was limited to the official abstract; the paper discloses that two authors previously worked or interned at the related advertising platform. We do not reproduce effect multipliers from the abstract.
