Review acquisition is usually treated as a post-launch operations problem. Amazon's September 29, 2026 announcement moves part of that work upstream. Review Requests lets eligible advertisers run campaigns through Amazon Ads Agent that invite recent verified purchasers to leave a rating or review while they are shopping on Amazon. A buyer can rate with one click and may add written feedback, images, or video.

This does not give a brand permission to manufacture reviews, script customer opinions, or optimize for five-star language. It changes the bottleneck. When the reminder is easier to deliver, the limiting factor becomes whether a real customer can quickly understand the product, observe what happened, and describe the experience honestly.

That makes proof design a launch-readiness decision for brands working with Chinese factories, creative teams, and localized product pages.

Start with the confirmed scope

Amazon Ads says Review Requests will enter open beta in the United States in late October 2026. It is available to registered sellers and vendors with an advertising console account in good standing and no active policy violations. A product is eligible while it has fewer than 1,000 total ratings and reviews; the campaign pauses automatically when that threshold is reached. Campaigns use Amazon Ads Agent or the Unified API and the bidding and budget controls used for other Amazon Ads campaigns.

Amazon also reported that products in its June 3–21, 2026 US closed beta received roughly three times as many ratings and reviews per week as they did before the test. That is Amazon's own limited test, not a forecast for every category, price point, or advertiser.

The policy boundary matters just as much as the launch detail. Amazon's inappropriate product reviews policy prohibits compensation for reviews, refund-for-review arrangements, asking only satisfied buyers, and package inserts that request positive reviews. Review Requests increases access to verified purchasers; it does not transfer control over what those purchasers say.

The new question: what can a buyer actually observe?

A star rating takes a tap. A useful review requires a customer to remember a situation: what they tried, what changed, and whether the result could help another buyer decide.

If the product's main value appears after three months, if installation conditions are unclear, or if the result cannot be distinguished without controlled testing, a more efficient prompt will not create better evidence. It may surface expectation and onboarding failures faster.

The opposite is also true. A portable projector can give a buyer an immediate focus, throw-distance, and connection moment. A smart lock can make door compatibility, installation, unlocking, and low-battery behavior observable on day one. A storage product can make size, capacity, and stability visible within minutes.

The brand should not tell the buyer what conclusion to reach. It can make legitimate product use easier to understand and document.

The ONCE Proof-Moment framework

Instead of commissioning “more videos,” ask which product promises a customer can verify during the first use. Run each important promise through four steps.

1. Convert the claim into a customer action

“Powerful,” “easy to install,” and “portable” are production dead ends. Rewrite each one as an action performed in a defined environment, with known accessories and a result that can be seen.

For a smart lock, “easy installation” might become: confirm door thickness, choose the correct hardware, secure the lock body, pair the app, and complete the first unlock. That tells a production team whether it needs a close-up, a dimension reference, a timer, or a troubleshooting insert.

2. Choose an observation that does not depend on enthusiasm

Different buyers may judge the same observation differently, and that is healthy. Dimensions, connection success, included parts, button response, cleaning steps, storage volume, and operating sound are more useful proof points than “premium” or “amazing.”

Durability, medical, health, safety, and precision-performance claims need controlled evidence, certification, or long-duration data. A lifestyle demonstration cannot replace those materials. First-use content should explain conditions, not impersonate a test lab.

3. Support first use without scripting the review

Support can include a listing demonstration, neutral packaging instructions, a short setup video, diagrams, or troubleshooting. Its job is to help customers use the product under the correct conditions. It must not request a positive rating, prewrite review language, offer a benefit, or route unhappy customers away from an honest review path.

Before filming, list the conditions most likely to produce confusion: required accessories, compatible sizes, wait times, charge state, operating environment, cleaning, and maintenance. Clarifying those details often creates more value than another enthusiastic talking-head variation.

4. Return themes to the product and content system

Teams can aggregate recurring questions, objections, and use cases to improve the listing, onboarding, or next production cycle. A customer’s specific words, images, or video are not automatically a brand asset. If the brand wants to quote a review, it must check the intended placement, preserve the meaning, confirm any required permissions, and follow Amazon's presentation rules. Amazon Stores, for example, requires customer reviews to be verifiable on the associated detail page and prohibits editing the review or showing the customer name.

The loop is “learn from real use and improve,” not “select praise and amplify it.”

Measure the cost of an approved proof moment

Review Requests uses bids and budgets, but the less visible cost is preparing truthful use evidence. Compare production options with this denominator:

Cost per approved proof moment = (product-fact mapping + production and post + instruction design + localization + compliance review + version maintenance) / number of completed and approved proof moments

“Approved” means the moment matches the product sold, states material conditions, avoids exaggeration, and can be used by the listing, support, or onboarding team. It does not mean the moment generated a positive review. Five-star volume should never become a production acceptance metric.

Imagine two equal-budget options. Option A delivers five emotionally similar short videos. Option B delivers three first-use proof moments, each with the action, observable cue, troubleshooting, horizontal and vertical coverage, and localized instructions. For a functional or installation-heavy product, Option B may cost more per clip but create a more reusable launch asset. For a highly aesthetic product with minimal setup, Option A may contribute more brand value. The right choice depends on the product risk and use path, not a universal modular-content doctrine.

Where this approach fits—and where it does not

Prioritize proof moments when:

  • first use happens within minutes or days and the outcome is directly observable;
  • setup, connection, dimensions, compatibility, or maintenance strongly affects satisfaction;
  • the product is new, review-light, and otherwise eligible for Review Requests;
  • the same demonstration can support the detail page, ads, customer service, and instructions;
  • unstated environment conditions commonly create disappointment.

Use more caution when:

  • the result takes months to judge;
  • the claim concerns health, medicine, safety, or another regulated outcome;
  • value is mainly subjective, such as taste, fragrance, or aesthetic preference;
  • the team has not fixed recurring returns, failed setup, or unclear instructions.

In those cases, improve the product facts, test materials, or customer experience before buying more review-prompt reach.

Add five fields to the next production brief

For every important product promise, write down:

  1. when the buyer can first verify it;
  2. what the buyer must do;
  3. what can be observed;
  4. which conditions could change the result;
  5. what source gives the brand the right to make the claim.

If the team cannot complete these fields, the problem is usually upstream of the camera. Pair the list with a product-fact source sheet so that a visual demonstration and the claim behind it stay aligned. This is also the missing production layer in an AI-assisted shopping environment: agents can summarize claims, but buyers still need real conditions under which those claims make sense.

Review Requests increases invitation density, not a brand's authority over customer opinion. The durable investment is to help real customers use the product correctly, notice what happened, and document the experience in their own words. Review readiness should therefore begin in the production brief, not as a post-launch recovery task.

Sources and limits

The production and procurement recommendations in this article are ONCE's analysis of public materials. They are not an Amazon commitment about production methods or review outcomes.

← TikTok Ads Are Leaving TikTok. Your Creative Needs a Placement Passport. Amazon Is Connecting Products to Local Services. Your Product Video Needs a Handoff Scene. →