How we pick what we recommend

We have no lab and own no test units. So instead of asking you to trust a testing claim, here is a method you could repeat yourself.

What we do

Start from the buyer's constraint, not the spec sheet. For this site the constraints are, in order: does it survive submersion at the depth this activity actually involves; does it stabilize on a moving, chopping platform; does the battery last the trip at the temperature the trip happens at; does it mount to the thing being mounted to.
Weight the specifications by use case, and publish the weighting. A camera that wins on resolution and loses on depth rating loses the kayaking roundup. The weighting table for each category is printed on the page, so you can disagree with it.
Read the primary sources. Manufacturer published specifications for depth ratings, sensor size, stabilization mode, battery runtime and card requirements — cited with a link and the date we read it, because manufacturers revise those pages.
Read the independent reviews that do test, and cite them by name and URL where our judgment rests on their findings. If we are leaning on someone else's test, we say whose.
Read owner reports for the failure modes that never appear in specifications — fogging, salt corrosion, cold-weather battery collapse, mount slippage — and label them as anecdotal rather than measured.
Score against a visible rubric. Four to six named criteria per category, each scored, with the weighted total shown. The rubric is printed on the page it applies to.
Say what we do not know. Where a specification is not published, the page says “not published” rather than filling the cell with an estimate.
Date everything, and revise when a named model is superseded.

What we explicitly do not do

We do not lab-test, bench-test or dive-test cameras, and we do not own test units. Nowhere on this site will you find “in our testing”, “our unit”, or an hour count attributed to our own use of a product.
We do not accept payment, product or copy from manufacturers in exchange for placement or ranking.
We do not rank by commission rate. Where a cheaper retailer is better for the reader, that is the link we use.
We do not publish a price we cannot verify at the moment you read it — which is why the buy buttons say “check price” rather than quoting a number.
We do not invent star ratings, review counts or testimonials. This site has none, and it has none because it is new, not because we are hiding them.
We do not put our editorial scores into structured data. The score you see on a page is our opinion, published as page content; marking it up as a product review rating would tell search engines something we have not earned the right to say.

The rubrics, in full

Each category is scored on its own criteria, weighted for lake and trail use. These same tables appear on the roundups they apply to.

Action cameras

30%Water survival: Rated depth without a case, door design, and what the manufacturer actually publishes.
25%Stabilization in chop: A kayak deck and a rooty trail shake in ways a handheld walk test never shows.
20%Battery in the cold: Published runtime, spare-battery cost, and whether it charges while recording.
15%Mount ecosystem: Finger mount, threaded, magnetic — and how much of it you have to re-buy.
10%Footage you will keep: Lens replaceability, low-light behavior, and how much cutting the footage needs.

Mounts

35%Hold under load: Vibration, chop and a wet hull are what actually break a mount.
25%Fit range: Rail width, bar diameter, helmet curve — the spec that decides whether it fits at all.
20%Water and salt tolerance: Stainless vs plated hardware, and what happens after a season of rinses.
12%Adjustability: Getting the angle right on the water, with cold hands.
8%Tether point: A mount with nowhere to clip a leash is a mount that loses cameras.

Memory cards

40%Sustained write speed: The number that decides whether high-bitrate video drops frames or stops.
25%Manufacturer compatibility list: GoPro, Insta360 and DJI each publish tested cards; that list beats any spec.
15%Capacity for the trip: How many hours at the bitrate you actually shoot.
12%Durability rating: Water, X-ray, temperature range — a card lives in a wet pocket.
8%Warranty: Cards fail; what happens next matters.

Batteries and power

35%Real runtime: Published minutes at the mode you actually record in, not the best case.
25%Cold-weather behavior: The stated operating range, and how fast capacity falls below it.
20%Charge while recording: Whether a long day is possible at all without stopping.
12%Bulk on the boat: Weight and size in a dry bag you also have to paddle with.
8%Cell quality signals: Certifications and warranty, since cell quality itself is never published.

How we cite

Claim	What we cite
Depth rating, IP rating, sensor size, card class, published runtime	The manufacturer specification page, linked, with the date we read it
Comparative performance — stabilization, low light, audio	The independent review we are relying on, by name and URL
Failure modes — fog, corrosion, overheating, cold battery drop	Labeled as owner-reported, never stated as measured fact
Prices	Never stated. The live price layer, or “check price on Amazon”
Anything we cannot source	Omitted, or written as “not published”

First-hand experience, and its limits

There is a distinction on this site that we take seriously. We have genuine first-hand experience of the use case — what a kayak deck does to a camera in chop, how much cold shortens a day, what salt does to a mount thread, where people actually lose cameras, why most family trip footage is unwatchable and what changes that. That experience shapes the orientation sections and the “what decides this category” blocks, and it is real.

We do not have first-hand experience of the products. Every product-level claim on this site traces to a manufacturer specification or a named independent test, and where the two disagree we say so. If you ever find a sentence here that implies we ran a named product through a named test, it is an error and we want to know about it.

Corrections and updates

Every page carries a published date and a last-updated date. The updated date moves only when the content genuinely changed — it is not the build timestamp, and a redeploy does not touch it.

We revise a page when:

a named model in it is superseded or discontinued;
a manufacturer revises a published specification we quoted;
a reader reports an error;
or six months pass, at which point every roundup and comparison gets checked whether or not anything has changed.

Found something wrong? Email janet.hoyo@yahoo.com or use the contact form. Material corrections are noted on the page itself, not quietly edited out.