How to Research a Product Review Video With AI

Write one decision question. “Is this microphone a sensible upgrade for a solo creator recording in an untreated room?” is stronger than “Is this microphone good?” because it identifies the user, environment, and alternative. Choose three to five criteria that could change the verdict, such as setup, performance under a repeatable condition, compatibility, ongoing cost, and return terms.

Write the test before opening product pages. Record the device or software version, settings, room or network conditions, comparison baseline, and pass or fail rule. This prevents a polished marketing claim from quietly becoming the test. It also gives AI a bounded job later: organize observations against declared criteria instead of inventing a verdict from general knowledge.

Collect the official product page, manual or specification sheet, current pricing page, warranty or return policy, and relevant release notes. Save the URL, page title, publication or checked date, and a precise locator such as a heading, table row, or PDF page. Prices, plan limits, and software features can change, so label them with a checked date instead of treating them as timeless facts.

Record the commercial relationship too: whether you bought the product, received it free, used an affiliate link, or were paid by the brand. The FTC says a material connection can include money, free or discounted products, employment, and personal or family relationships. Its creator guidance also says an endorser should not describe an experience they have not actually had.

Ask AI to extract candidate claims from one source at a time. Do not ask it to merge everything into a review yet. For each candidate, record a claim ID, exact wording, source and locator, checked date, classification, required test, observed result, status, and planned place in the script. Use accepted, qualified, disputed, unsupported, and stale as explicit statuses.

“Battery capacity is 5,000 mAh” may be a documented specification. “Lasts all day” is a performance claim whose meaning depends on use. “It lasted seven hours in my test” is an observation that needs the test conditions. “It is enough for most travelers” is an inference. Those statements should remain separate even when a smooth summary would combine them.

Prioritize consequential claims instead of every bullet on the sales page. If speed, battery life, image quality, compatibility, or total price determines the recommendation, test or verify those first. Capture the setup, raw result, failure, retest, and any condition that limits the finding. Keep clips or screenshots that let an editor reconstruct what happened.

Compare the result with the precise claim. A claim may be supported, contradicted, or simply untested. “I could not reproduce this under these conditions” is more useful than declaring the company wrong when the method covers one setup. If a feature cannot be tested, attribute it directly to the manufacturer and keep it out of the personal verdict.

Open with the viewer's decision and the short answer. Explain what you tested, the conditions, and what you did not test. Present the two or three findings that most affect the choice, name a limitation or unresolved claim, and end with best-when and poor-fit-when guidance. Keep claim IDs in the working script until the final fact check.

Separate official facts from observations in narration and on screen. Add descriptive source links in the description, but do not hide a commercial disclosure there. The FTC says a video endorsement disclosure should appear in the video and be hard to miss; it notes that audio and on-screen presentation together are more noticeable than either alone. Its 2023 Endorsement Guides update also warns that a platform's built-in tool may not be sufficient by itself.

Read the final script from the verdict backward. For every product fact, price, quotation, or performance statement, open the source or test record and confirm the wording, locator, date, and scope. Recheck dynamic facts on publication day and remove any claim that became broader during editing.

YouTube requires creators with paid product placements, sponsorships, endorsements, or other commercial relationships to use its paid-promotion setting. Review its current instructions before upload, verify that the spoken disclosure, on-screen disclosure, setting, and description agree, and follow any additional law or contract that applies. Another editor should be able to tell what the manufacturer documented, what you observed, under which conditions, and what relationship could affect the recommendation.