Methodology

How drywallDB treats data

Construction decisions get audited — by clients, approvals processes and, occasionally, courtrooms. So drywallDB is built to be checked: every value on screen belongs to a stated class, every equivalence judgement cites the versioned rule that produced it, and the boundaries of what the tool will and won’t claim are written down here, in public. This page is the working behind the product.

The three classes of data

Every value drywallDB shows belongs to exactly one of three classes. The same three classes are encoded in our terms of service — this page is the plain-English version of a binding commitment, not marketing gloss.

Declared

Taken verbatim from a cited manufacturer document — a product data sheet, a declaration of performance, or the manufacturer’s own published listings. Each declared value names its source document and revision, and the source documents are archived, so a citation points at the exact edition the value came from, not at a moving web page. Values are reviewed against the source line by line before they enter the catalogue. Where a manufacturer’s own documents disagree with each other, we query the manufacturer and record the answer rather than silently picking a side.

Calculated — always marked *

drywallDB arithmetic over declared inputs, and nothing else. Every calculated value carries the asterisk mark, everywhere it appears — results table, comparisons, data sheets and printed documents alike. The full list of what drywallDB calculates is deliberately short:

• Board weight (kg) — a declared kg/m² multiplied by the board’s nominal area, shown only where the manufacturer declares no per-board weight.
• Board area (m²) — nominal width × length.
• kg/m² where none is declared — a declared per-board weight divided by nominal area, so weight filters and sorting still work for boards whose sheets declare only a per-board figure.
• Weight deltas in comparisons — candidate minus benchmark, the cost proxy an estimator prices.
• Per-board rate figures — a pack or per-m² rate you enter converted to a per-board figure. These are private to your account and never published.

Calculated values inherit the declaration basis and nominal character of their inputs. Manufacturers state weight on different bases — one brand declares a minimum, another a typical mass — and a figure derived from a minimum is still a minimum. The mark is the honesty mechanism: you can always tell our arithmetic from a manufacturer’s declaration.

Indicative

Estimates identified as such. The class exists in our terms and our data model, and it is currently empty — nothing in the catalogue today is indicative. If estimated data ever ships (for example, computed build-up performance), it will be labelled as this third class, clearly separated from cited data, and never presented as certified performance.

The equivalence rules — versioned, published, explainable

Cross-brand equivalence is drywallDB’s signature judgement, so it is the last place a hidden heuristic belongs. Matching runs on a fixed, published rule set — currently ≈ rules v1 — and every comparison and substitution document states the version that assessed it. If the rules ever change, the version number changes with them: a document generated under v1 stays attributable to v1. The rules in full:

1. Same thickness, different brand. Candidates must match the benchmark board’s thickness exactly — a 15mm board is never proposed for a 12.5mm spec — and come from another manufacturer.
2. The declaration must be covered. Every performance class the benchmark board declares to EN 520 (Type F fire, D density, I impact, R strength…) must also appear on the candidate. Missing one rules it out.
3. Type A never decides anything. It’s the baseline every plasterboard meets; some manufacturers print it, some don’t. Treating it as a requirement would split identical boards on house style, so it’s read as a labelling convention.
4. Moisture may be equal or better — never worse. H classes are ordered (H1 best, then H2, then H3). A candidate can match or exceed the benchmark’s class; when it exceeds, the upgrade is stated on its column rather than silently absorbed.
5. Different forms never swap. A plaster baseboard (Type P), a foil-backed vapour board (EN 14190) and a glass-mat wet-area board (GM) are different products, not upgrades — these must match exactly, in both directions.
6. Best format wins. Among qualifying boards from a brand, the one closest to the benchmark board’s size and edge is offered.

Two disclosures ride with every assessment. Reaction to fire is never a pass/fail criterion — where the classes differ, the difference is shown for your judgement, not silently gated. And boards that cover the spec and more are listed separately as over-specified alternatives with their extra classes named, because over-performing board usually costs more — they are never mixed in as if they were like-for-like. You can see all of this working on the annotated example comparison.

What drywallDB does not do

The boundaries matter as much as the features. Each of these is a deliberate refusal, stated in the product wherever it is relevant:

• No fire-resistance minutes for boards. Reaction to fire (e.g. A2-s1, d0 to EN 13501-1) is a board property and is shown, cited. Fire resistance in minutes belongs to a tested system — a specific build-up tested with specific boards — and no manufacturer certifies minutes for a board alone. We won’t print data we can’t cite, so the minutes stay in the manufacturer’s system literature, and every fire-related comparison carries that caveat.
• No performance figures for brand-agnostic build-ups. Fire and acoustic performance is certified per manufacturer-tested assembly. Generic “any 12.5mm board on 70mm studs” composites have no citable source, so drywallDB does not display them — even though shipping them would be easy and look impressive.
• No certification. Equivalence assessments are an aid to professional judgement, not a certification. Board-for-board substitution within a fire-rated build-up is not automatically valid, and drywallDB says so wherever fire classifications are involved. Responsibility for specification decisions remains with you and your project’s approvals process.
• No “proving” a spec’s values. When you check a specification’s claimed values, drywallDB confirms them against the manufacturer’s current published declaration — it does not prove them true. A declaration is the manufacturer’s claim; drywallDB tells you what that claim currently is, and cites where it is made.
• No invented conversions for withdrawn standards. No official mapping exists from BS 1230 board types or BS 476 “Class 0” to today’s EN classes. drywallDB explains what replaced each withdrawn standard and when — and where no official conversion exists, it says exactly that instead of publishing a table no standards body has ever issued.
• No market prices. drywallDB never publishes or aggregates market prices. The rate book holds what your own merchants quote you, privately — rates never appear in shared links, outward documents or any aggregate. Priced comparisons are priced at your figures, not a list price nobody pays.

Where this is binding

The three data classes and the not-a-certification stance are not just principles — they are written into the terms of service that govern every account. Manufacturers revise their literature; always verify against current manufacturer documents before purchasing or certifying.