Benchmark · August 2026

A specification score cannot know your roof, your grid, or an installer's own record with a product. We have written that limit down.

Every score on this site comes from the same published rubric applied without exception: a specification, a certification, a warranty term, run through the same arithmetic on every product. It was never built to know your roof, your grid connection, or the years a local installer has spent watching one brand outperform another in practice. That is a structural limit, not a footnote, and it now has its own line in our methodology.

Why this needed saying

The rubric already excludes what it cannot verify. This is the same principle, aimed at the reader's own situation.

Price, BloombergNEF's bankability tier, PV ModuleTech's reliability rankings and regional certification are already excluded from every score here, each for a stated reason: they change too fast, or rest on data this site cannot independently verify, or vary by market in a way a global rubric should not penalise. Local installer judgement belongs on the same list, for the same reason. It is real evidence. This benchmark has no way to see it.

What a specification cannot capture

Four things that never appear on a datasheet.

Site-specific, not published

What a specification score cannot see

1. 01
   Roof and shading

   Orientation, pitch, partial shading and roof material affect real output in ways no rubric criterion measures.

2. 02
   Grid conditions

   Voltage rise, export limits and local network rules vary by feeder, not just by country. An installer working the same substations for years has seen what actually trips a limit, and what does not.

3. 03
   Climate-specific durability

   This site has published laboratory results showing one commercially marketed encapsulant lost roughly 55% relative power after 1,000 hours of damp heat, against about 11% for standard EVA in the same test. A datasheet carries neither figure. An installer who has opened up failed systems in a humid climate has seen the difference directly.

4. 04
   Service and warranty record

   This site's own Verified reliability criterion is scored at brand level, because third-party testing programs assess representative models, not every SKU. An installer's own callout history with a brand is a finer-grained, site-specific version of the same signal, for the specific product in front of them.

What this does and does not mean

A lower-ranked product is not a wrong recommendation.

This is not an instruction to distrust the rubric, and it is not a licence for any recommendation to override a sourced fact with an unsourced one. A retailer or installer favouring a particular product still has to answer the same question this site already asks of itself: what is actually published, and can it be checked. What changes is narrower, and more honest. Where a trusted local installer recommends a specific product against this benchmark's ranking, on the strength of site conditions and a track record this benchmark cannot see, that is not a contradiction of the score. It is a different kind of evidence, and a reader is entitled to weigh both.

This is a standing methodology position, not a one-time note. Read the full statement, alongside every other thing this benchmark declines to score, on the methodology page.

Sourcesthis site's own published methodology

• review.solar methodology, "What we do not score": price, BloombergNEF bankability tier, PV ModuleTech reliability rankings, regional availability and certification, and local installer judgement. See the full list.
• Damp-heat encapsulant result (roughly 55% relative power loss against about 11% for EVA, 1,000 hours) from this site's own third-party assessment coverage: Three names on a spec sheet. Only one of them is about quality.
• Verified reliability criterion methodology: third-party reliability testing (PVEL, Kiwa/DNV, RETC) is awarded at manufacturer/BOM level, not per individual SKU, as recorded in this site's own scoring provenance.