How We Review Sleep Apps & Tools
This page is our promise. Every review on Tested Sleep carries a visible evidence label, and if a page cannot meet an evidence standard, we say so instead of pretending.
1. Three levels of evidence
Not every review is the same kind of review. Before you trust a conclusion, you should know exactly what kind of work produced it. Every review on this site is labeled with one of three evidence levels:
Level 1 — Documented Review
Built from official product information (product pages, help docs, pricing, privacy policy, update history), a structured analysis of publicly available user feedback, and human editorial review. It does not require installing or continuously using the app, and it must not claim experienced accuracy.
A Documented Review may evaluate feature completeness, patterns in user feedback, stability reports, privacy transparency, pricing and value, and update or support status. It must not present as tested fact: sleep-stage accuracy, fall-asleep or wake detection error, real battery use, microphone recognition, actual comfort, or the reviewer's own long-term experience.
Level 2 — Hands-on Review
John Lau or a clearly named tester actually installs and uses the app. The device, OS, app version, test dates, and main actions are recorded. If fewer than 14 nights are completed, it cannot be labeled 14-Night Tested.
Level 3 — 14-Night Tested Review
The same primary tester uses the app continuously for at least 14 nights. Device, OS, app version, and test setup are stated. Night-by-night logs and result evidence are kept. Only at this level may the label "14-Night Tested" be used.
Levels are not inverted. A Documented Review is a different evidence type, not a lower-quality fake review.
2. Comment sampling principles
For each Documented Review we publish the platform(s) sampled, the scrape or query date, the time range, and the valid sample size. We collect positive, neutral, and negative comments. We exclude empty, duplicate, and clearly irrelevant comments. A star average is not treated as accuracy, and a single comment is never generalized into a universal fact. We separate official facts, user reports, and editorial inference. We cite a few necessary short quotes; the body uses summary and paraphrase. Every important conclusion has a traceable source.
3. Source priority
Official product materials first; then structured public user feedback; then reputable published sources. Claims are cross-checked across sources before publication.
4. The role of AI
AI assists with source discovery, organization, classification, and draft preparation. AI does not use, sleep-test, or experience the apps, and is never described as a real product tester.
5. John Lau's editorial responsibility
John Lau is the owner and editor. He reviews the evidence and approves every published conclusion. Editorial review and evidence verification: John Lau.
6. Scoring and verdicts
A Documented Review (Level 1) has no numeric score and no final verdict. Scores and verdicts are assigned only at Level 2 (hands-on) or Level 3 (14-night), with written evidence for each scored dimension. Missing evidence blocks a score rather than lowering it.
7. What we never claim
- That an app diagnoses, treats, or cures any medical condition.
- That nightstand tracking measures sleep stages with medical accuracy.
- That something was personally tested when it was not, or that AI tested it.
- Experienced accuracy (sleep-stage, detection error, battery, microphone, comfort) from a Documented Review.
8. Corrections and updates
Prices, features, and apps change. Every final review is dated, and we update or re-review when a major version or pricing change affects the conclusion. Documented Reviews note their research period and last-checked date.