How We Choose and Review
Everything we publish is built from published sources — manufacturer records, retailer listings, user reviews and named professional references. We do not run a testing laboratory, and nothing on this site claims hands-on measurements.
No testing lab. AIBUYLAB does not run a product-testing laboratory. Our articles never claim hands-on measurements, months-long test periods or controlled comparisons. When we rank a product first, that is an editor's judgment based on the evidence below — not a bench-test result.
What we use. Every article is built from three kinds of source: the manufacturer's published specifications and product records; official retailer product pages (features, availability, current offers); and user reviews and public discussion we can point to. Where sources disagree, we prefer the manufacturer record for specifications and the retailer page for current offers, and we say so in the article.
How articles are produced. Coverage is drafted with AI assistance against those listed sources, then reviewed and edited by a section editor before publication. Articles that fail our quality gates are held back rather than published. Every article shows a byline and an updated date — and a reviewer line whenever our editorial roster has a second editor; best lists also keep an update log.
Pricing and links. Any price printed in an article is the public list price recorded on the verification date shown on that page — list prices change, so the retailer's page is always the source of truth for what you pay today. Where a live price feed is available, the page loads it instead of an old number. Purchase buttons are plain, direct links to the retailer's product page — AIBUYLAB currently has no affiliate relationships and earns no commission from any purchase.
Corrections. When a specification, ranking or price detail is found to be wrong, we update the article and add an entry to its update log. If you spot something that does not add up, tell us — we would rather correct it than leave it.
How we produce an article
Every article follows the same five steps. Nothing is published until each step is done — an article that fails a gate is held back rather than fixed in public.
- Find the sourcesWe start by collecting what can be cited: the manufacturer's published product record and specifications, the official retailer listing, user reviews and public discussion, and named professional or veterinary references. A claim with no source does not go in the article.
- Draft against the sourcesCoverage is drafted with AI assistance against those listed sources. Every specification, price and review count in the draft must trace to a source on the list — anything that cannot be traced is cut or marked as unverified.
- Section editor reviewA section editor reads the draft against the sources, checks the numbers a reader would check, and removes any claim that overstates the evidence — hands-on testing, expert endorsements, or a 'best' pick without a published score behind it.
- Quality gatesAutomated gates check what can be checked mechanically: titles match the number of products, prices carry a verification date, dimension scores carry a data-source label, and no banned testing claim appears. Failing articles are blocked, not patched.
- Publish with a byline and a datePublished articles carry the writer's byline, an updated date, and a reviewer line whenever our editorial roster has a second editor. Best lists also keep an update log so changes are visible.
How we score: the AI Buy Lab Score
Products are scored on a 10-point weighted index called the AI Buy Lab Score. Each category has 5 to 6 scoring dimensions — for example cleaning performance, noise and capacity for litter boxes — and each dimension carries a published weight. A product's final score is the weighted average of its dimension scores, and the formula is public below.
Every dimension score carries one of three data-source labels: manufacturer specifications, aggregated user reviews, or editorial assessment. No score is a hands-on measurement — a score summarizes published evidence, and the label tells you which kind. Dimension weights for every category are published below, and each product page shows its own dimension scores with their sources.
How the score is calculated: final score = the sum of each dimension score multiplied by its weight, divided by the sum of all weights, on a 0 to 10 scale, rounded to one decimal place.
Every dimension score carries one of three data source labels: manufacturer specifications (from the maker’s published product record), aggregated user reviews (public ratings and review counts on the maker’s or retailer’s page), or editorial assessment (our judgment from those two kinds of evidence).
AIBUYLAB does not run a product-testing laboratory — no score on this site is a hands-on measurement.
| Litter Robots | Cleaning performance (25), Noise (15), Capacity & multi-cat fit (20), Consumables cost (10), Ease of cleaning (15), Smart features (15) |
|---|---|
| Automatic Feeders | Portion control (25), Food compatibility (15), Scheduling & app (15), Ease of cleaning (15), Power & backup (10), Value (20) |
| Water Fountains | Filtration (25), Capacity (15), Quietness (15), Ease of cleaning (15), Power & placement (10), Value (20) |
| GPS Trackers | Location accuracy (25), Battery life (20), Coverage (15), Durability (10), Subscription cost (15), App & alerts (15) |
| Pet Cameras | Video quality (25), Two-way audio (15), Night vision (15), Smart features (15), Storage & subscription (10), Value (20) |
How we keep coverage current
Coverage is dated, not timeless. Every article shows an updated date; best lists keep an update log with one entry per material change. Before each deployment we re-verify that every purchase link is reachable, and any price printed on a page carries the verification date it was recorded on — the retailer's page is always the source of truth for what you pay today.
When a product is discontinued or superseded by a newer model, we note it in the coverage and in the affected best lists, and re-check the remaining picks. When a specification, ranking or price detail is found to be wrong, we correct the article and add an entry to its update log rather than silently editing it.
Limitations
We would rather you know exactly what this site is and is not. These limitations apply to every article, not just the small print:
- No testing lab. We do not run hands-on measurements, durability tests or controlled comparisons — no score here is a measurement, and rankings are editorial judgments from published evidence.
- Scores summarize other people's data. Dimension scores labelled manufacturer or user-reviews come from the maker's published record and public review aggregates; editorial scores are our judgment from that evidence. Where evidence is thin, the score is thin too.
- Prices move. Any price shown was recorded on the verification date printed with it — check the retailer's page before buying. Where a live price feed is available the page loads it instead of an old number.
- Pet health content is general information. Health articles paraphrase published veterinary sources and are not veterinary advice — if your pet is unwell, consult your veterinarian.
- A one-editor newsroom. Our editorial roster currently has one editor, so articles carry a byline without a second-editor reviewer line; the reviewer line returns automatically when the roster grows. Drafts are AI-assisted against listed sources and reviewed by that editor before publication.