Editorial policy

Sourcing standards

Every factual claim about accuracy, specification or performance is either linked to its source, or explicitly marked as unavailable. There is no third option, and no claim gets to sit on the page unsupported because it sounds reasonable.

We read sources in full before citing them

Never from a search snippet, and never from a press release. This is not a formality — it catches real discrepancies. Oura's press release about the 2024 Brigham and Women's study describes it as finding 79% four-stage sleep agreement; the paper reports 76.3%. We use the paper, and we tell you the difference exists.

Where we could not read a source, we list it as unread and quote nothing from it. Two relevant papers were behind publisher blocks on 24 July 2026 and appear in our source lists marked exactly that way. Presenting them as support would have been easy and nobody would have checked. That is precisely why we do not.

We date every study

A 2020 validation of a 2019 sensor does not describe a 2026 device. Every source carries its year, and where a study tested a superseded model we say which model it tested — repeatedly, every time we cite it, not once in a footnote.

The banned-claims list

These never appear on this site, in any form, on any page. This is an internal rule that binds every piece of copy we publish:

• "We tested", "we wore", "in our testing", "our test unit", "after 200 miles" — we own none of these devices
• "Expert", "specialist", "veteran", "seasoned", "long-time", "X years of" — credential and tenure claims we cannot back
• CPT, CSCS, MD, "sports scientist", "medically reviewed by", "our expert panel" — credentials nobody here holds and staff who do not exist
• Any invented price, rating, review count or testimonial — there are no testimonials on this site because we have no customers, and inventing them would be catastrophic rather than merely wrong
• Unqualified superlatives— "the best watch, period" is unfalsifiable and therefore meaningless
• Fake urgency— no countdowns, no "only 2 left", no invented sale framing

Corrections policy

Corrections are logged visibly on the page, not edited in silently. If we get something wrong and fix it, the page says so and says when. A site that quietly rewrites its errors is asking to be trusted on exactly the thing it has just demonstrated it should not be.

How it works:

1. Report it via the contact form, which leads with "report an error" as its first option, or email info@pulseandpath.com. A link to the source helps most.
2. We check the source. If you are right, we fix it — normally within a few days for factual errors, immediately for anything that could mislead a purchase.
3. The correction is noted on the page with the date and what changed, and the last-updated date moves.

We are especially interested in one category of correction: a validation study we have missed. Our accuracy pages say "no validation study exists that we could find" in a number of places, and each of those is a claim you should try hardest to falsify. If one exists, that is a genuine improvement to the site rather than an embarrassment.

Update cadence

Last-updated dates are real.Bumping a date without changing content is a small dishonesty that would undermine everything else here, so we do not do it. The "last reviewed" date on trust pages is a hand-maintained constant rather than a build timestamp, for exactly that reason — otherwise it would advance on every deploy without anyone having read anything.

Independence

• No sponsored posts. No article on this site has been paid for, and none ever will be.
• No paid placements. No manufacturer can buy a position in a ranking.
• No free units. We accept no review samples — which is easy, since we do not review by testing.
• No manufacturer preview. Nobody sees a page before it publishes.
• Commission does not affect rank. The rubric decides, and the rubric is published with its weights so the claim is checkable rather than a promise.

In practice: where the best answer for a buyer is a product we earn nothing on, we say so and link it anyway, marked as non-earning. And several devices we could recommend enthusiastically for commission are instead ranked with an explicit note that nobody has ever measured them.

AI disclosure

Content here is researched and drafted with AI assistance, then fact-checked against primary sources before publication. Every study cited was fetched and read; every specification was taken from a manufacturer page; every price comes from a live retailer API. No figure on this site was generated from a model's recollection, and where a source could not be retrieved, nothing was quoted from it.

We mention this because the alternative — saying nothing and letting readers assume — is the same species of omission as an unstated affiliate relationship.