How we test and score products

1. How a product gets reviewed

We don't review products because a brand asked. We review the products our readers are searching for, the ones our newsletter subscribers ask about, and the ones we think will matter in the next 12 months. The selection criteria, in plain language:

• Search demand. The product is being searched for in numbers — Google, YouTube, Reddit, Quora.
• Buyer relevance. The product is something a US/EU consumer can actually obtain, now or in a defined near future.
• Category importance. The product is shaping how a category is being talked about, even if it isn't yet shipping at scale.

We don't review products that are vapor (no shipping date, no public spec), products restricted to non-consumer channels, or products where we can't get hands-on access in a reasonable timeframe.

2. How we get the product

We pay for review units whenever possible. When a manufacturer ships us a unit at no cost, we disclose it inline at the top of the review and the unit goes back to the manufacturer at the end of testing — or, if they decline its return, it's noted publicly in the review.

For products we can't physically obtain (e.g. China-only PettiChat), we say so up front and limit the verdict to what's verifiable from the company's published materials, third-party reviewer footage, and regulatory filings. We never pretend to have tested something we haven't.

3. What we test

We test the things we'd care about if we were the buyer. For an AI pet collar that's six dimensions:

• Build & wear comfort. Weight, ergonomics, chafing, water resistance under realistic conditions. Minimum 14 days continuous wear on a real dog or cat before a comfort verdict lands.
• Stated function. What the company says the product does — does it actually do that, at the stated rate, in our conditions. We test against the spec sheet, not the marketing.
• App experience. First-run, ongoing daily use, settings discoverability, notification quality, dark patterns.
• Privacy posture. What data is collected, where it goes, who has access, what jurisdiction applies. We read the privacy policy and the data-processing addendum.
• Battery and reliability. Claimed vs measured battery life, charging behaviour, connection reliability over a multi-week window.
• Customer support. We file a real support ticket and time the response. We try the warranty process if the unit misbehaves.

4. How we score

The verdict — buy, wait, or skip — is editorial, not algorithmic. But it's grounded in a deterministic scoring rubric so readers can compare verdicts across reviews:

• Buy. The product does what it claims, the price is fair for what's delivered, the privacy/support posture is acceptable, and no near-term alternative is materially better. We issue a buy when we'd buy the product ourselves.
• Wait. The product is real but flawed, or the category is moving so fast that a near-term ship from a competitor is likely to change the calculus. We issue a wait when we don't think you should regret buying later.
• Skip. The product underdelivers vs its marketing in a way that won't be fixed by a software update, the price is wrong for what's delivered, or the privacy / support / company posture is unacceptable. We issue a skip when a buyer would be worse off than not buying at all.

Every verdict carries a one-line reason, in plain language, at the top of the review. We don't bury the conclusion.

5. What we don't do

• No sponsored content. Brands cannot pay us to review their product, and they cannot pay us to influence a verdict. Newsletter sponsorships are clearly labelled and never touch editorial.
• No quid pro quo with manufacturers. We don't show verdicts to manufacturers in advance. We don't soften verdicts in exchange for future review units.
• No fake first-person. If we haven't tested something hands-on, the review says so. We don't write "I noticed the battery drains quickly" if we don't have the device.
• No private affiliate sweeteners. The affiliate programs we use — Amazon Associates, direct manufacturer programs — are listed publicly on our disclosure page. Commission rates are public information.

6. Re-reviewing

Categories move. A "wait" today may become a "buy" when the competitor ships, or a "skip" if a privacy incident lands. We re-review when:

• The product ships a major firmware or app update.
• A competitor materially changes the buying landscape.
• A privacy, security, or business-model change occurs.
• It's been 12 months since the last verdict and the product is still on shelf.

When a verdict changes, we update the post with a dated note at the top and ping subscribers in the Weekly Buyer's Brief.

7. Reporting an error

If you have a primary source that contradicts something in a review, we want to hear from you. Email editor@aianimalcollar.com with the affected URL and your source. Corrections are visible and dated.

This page is versioned

When we change anything material on this page, we'll note it in the Weekly Buyer's Brief and date the change here. Last updated: May 27, 2026.