Ecommerce Ad Research Framework: From Product Truth to Video

An ecommerce ad does not become useful because an AI model produced fluent copy. It becomes useful when the advertising argument can be traced back to product facts, market evidence, and decisions the advertiser has reviewed.

This ecommerce ad research framework explains the chain MotionSKU uses before video generation. The goal is not to predict a winning ad. The goal is to find a focused, supportable argument worth testing before production credits are spent.

The complete research chain

The framework moves through eight connected layers.

1. Product Truth establishes what the supplied product URL, images, and user corrections can support.
2. Voice of Customer, or VOC, captures situations, frustrations, alternatives, objections, and desired outcomes in customer language.
3. Buyer Profile narrows the research to one buying situation, awareness stage, and trigger moment.
4. Mass Desire identifies the broad outcome the audience already wants.
5. UMP explains why the current problem or friction continues.
6. UMS explains how the product changes that situation.
7. Offer Brief defines the promise, proof, offer facts, objections, CTA, and claim boundaries.
8. Necessary Beliefs connect the opening Hook to the product proof and the next step.

Each layer reduces ambiguity. Skipping directly from product page to script forces the model to invent the missing strategy inside the copy.

Step 1: establish Product Truth

Begin with the product source. A public product page may provide a title, images, structured product data, specifications, price, offer details, usage, and product claims. Uploaded images may prove shape, color, packaging, included parts, texture, or a visible product action.

Separate every input into an evidence state.

• Product fact means the information appears in the supplied source or is directly visible.
• Market source means the idea was observed in a public review, forum, competitor page, or ad library with source context.
• Model inference means the AI proposed the idea from available context.
• User confirmed means the advertiser corrected or explicitly approved it.
• Missing evidence means the information is important but not supplied or verified.
• Generated example means synthetic copy or media demonstrates a method without claiming customer performance.

This distinction matters because a plausible statement is not automatically a usable advertising claim.

Step 2: research customer language

VOC research should look for repeated patterns, not isolated dramatic sentences. Useful categories include the moment the problem becomes noticeable, the cost of leaving it unresolved, previous attempts, objections, desired outcomes, and the words buyers use to compare alternatives.

Record the source type and paraphrase when quotation rights or context are uncertain. Do not turn model-generated language into a customer quote.

Read the full Voice of Customer research guide for a practical collection and synthesis method.

Step 3: define the Buyer Profile

The Buyer Profile is not a demographic collage. It is the person and situation relevant to one advertising argument.

A useful profile answers five questions.

• What is happening when the product becomes relevant?
• What does this buyer already know about the problem and possible solutions?
• What have they tried or accepted as the current alternative?
• What objection could stop the next step?
• What outcome do they want in plain language?

This profile guides the script. It is different from an AI Avatar, which is only the visible presenter used during production.

Step 4: choose one Mass Desire

Mass Desire is the broad outcome the market already wants. It is not simply the loudest pain point.

For example, a buyer may complain about a slow morning routine. The deeper desire could be to leave home feeling prepared without adding more steps. The script should use the pain to create relevance, then move toward the desired outcome.

MotionSKU presents three one-line Mass Desire options so the user can make this decision without reading the full dossier. The supporting VOC, Buyer Profile, mechanism, proof, and objection remain available under review.

Step 5: articulate UMP and UMS

UMP is the Unique Mechanism of the Problem. It explains why the unwanted situation continues despite previous attempts.

UMS is the Unique Mechanism of the Solution. It explains how the product creates a different result in an understandable, supportable way.

A mechanism does not need scientific language. For a physical product, it may be a design choice, a simpler action, a placement difference, or a repeatable use pattern. If the source cannot support a medical or technical explanation, the mechanism must stay with what the evidence shows.

Step 6: write the Offer Brief

The Offer Brief turns research into a creative boundary. It should contain the approved audience situation, promise, product proof, UMP, UMS, objection handling, offer facts, CTA, and excluded claims.

This document is deliberately shorter than the research dossier. It helps every later model work from the same approved decision set.

Use the ecommerce Offer Brief template to build one.

Step 7: list the Necessary Beliefs

Necessary Beliefs are the small set of ideas a viewer must accept before the product and CTA feel reasonable.

A typical sequence might be:

1. This situation is relevant to me.
2. The usual approach leaves a specific friction unresolved.
3. This product changes that friction through a visible or understandable mechanism.
4. The available proof is enough to consider the next step.

These are planning statements. They should not be presented as customer quotes or universal beliefs.

Step 8: generate three complete scripts

The three options should differ in the Hook while keeping the approved strategic spine visible. One Hook may begin with the trigger moment, one with the desired outcome, and one with the product action.

Each option should show the complete 15-second script, including the middle argument, product proof, and CTA. Choosing only an isolated opening line makes it difficult to judge whether the Hook can support the rest of the ad.

From script to video plan

The approved script becomes a timed video plan. For the MotionSKU MVP, the default is a 9:16 video of no more than 15 seconds.

The plan checks original product references, one optional 9:16 Master Visual, dialogue, on-screen text, timing, estimated credits, and explicit generation intent. The Master Visual helps coordinate character, setting, product placement, and visual tone. It does not replace the original product images.

Research boundaries

This framework cannot guarantee conversions or return on ad spend. It should never invent customer testimonials, experts, test results, prices, discounts, guarantees, certifications, or performance outcomes. Repeated competitor advertising can be a research signal, but it is not proof that the creative converts.

The result of good research is not certainty. It is a clearer reason to test one argument instead of generating random variations.

Frequently asked questions

Do I need a large research team?

No. A language model can collect and organize a first research pass, while the user reviews sources, assumptions, and strategic choices.

Can Product Truth come from images alone?

Yes, but the evidence is narrower. Images can establish visible details while benefits, usage, offers, and performance claims may still require user input.

Should all research appear in the final ad?

No. Research creates options and boundaries. The final 15-second ad should make one focused argument.

Research your product before rendering

Use the ecommerce video ad generator to confirm Product Truth, compare Mass Desire options, review three complete scripts, and approve the video plan before generation.