Methodology
How we choose what to recommend
Including a full account of what we cannot claim, which is the part most sites in this category leave out.
The honest version, first
Deck & Dusk does not run a testing facility. We have not weathered these products for five years, we have not measured wind loads, and we have not put fabric samples in a xenon-arc chamber. Any site telling you it has done those things for a hundred-product roundup is telling you something that is not true.
What we do instead is apply the same criteria consistently, from evidence you can check. Every product recommendation on this site carries this sentence:
Selected on published specification, warranty terms and owner-reported outcomes; not personally tested.
That is the single clearest thing separating this site from the incumbent set, several of whom describe months of outdoor exposure and hundreds of hours of testing for product ranges nobody could plausibly have tested.
The five criteria
Everything is scored out of ten on the same five things, and the same reasoning is applied across every product in a category.
| Criterion | What we look at | Where the evidence comes from |
|---|---|---|
| Fade resistance | Whether color is compounded or solution-dyed into the material, or applied to its surface | Manufacturer material statements and published fabric warranties |
| Rust and corrosion | Frame metal, fastener grade, and whether any corrosion path exists at all | Specification listings, and metallurgical references for fastener grades |
| Wind and stability | Weight, footprint, center of mass and sail area | Published dimensions and shipping weights; observable geometry |
| Build and hardware | Joinery type, fastener material, and whether parts remain available | Specifications, assembly documentation, retailer parts listings |
| Warranty strength | Duration, and crucially whether the terms name specific failure modes | The published warranty document itself |
Scores are editorial assessments derived from these inputs. They are not measurements and are never presented as such.
What we never claim
- Laboratory or instrumented testing. There is no lab. We will not describe one.
- That we have used every product. Where a product has genuinely been used first-hand, that is stated per product. Everywhere else it is not.
- Unquantifiable effort. No "months of exposure" or "hundreds of hours" — phrases that sound like evidence and are not.
- A score as a measurement. Our numbers are judgments made explicit, not data.
- Customer reviews or ratings we do not have. There are none on this site because there are none to show.
How sourcing works
Every attributed claim on this site resolves to an entry in a citation registry that records the publisher, the document title, the URL and the date it was last checked. A claim without a matching source entry does not get published — that is a build constraint here, not a policy someone has to remember.
Pages that state lifespan, warranty or specification figures carry a Sources section at the bottom listing exactly what was used and what it was used for.
How products get on a list
- The product must be real and currently sold. We do not list discontinued models to fill a category.
- It must fit the page honestly. If no product genuinely suits a page, the page runs without picks. Safety and troubleshooting content is legitimately unmonetized.
- It is scored on the five criteria. Same reasoning for every product in the category.
- The trade-off is named. Every entry lists what the product gives up. A pick with no downside listed is a pick nobody looked at properly.
- Ranking is buyer-first. Where a cheaper option is genuinely better for the buyer, it ranks higher, regardless of what it pays.
Updates and corrections
Pages carry a published date and a last-updated date, both visible at the top. We update when a warranty changes, a product is discontinued, or a source we relied on is revised or withdrawn.
If something here is wrong, please tell us — the contact page reaches the person who wrote it. Corrections are made on the page itself rather than quietly in a changelog.