Published so that readers can hold us to them, and so that a clinic can judge whether an instrument deserves the time it asks for.
How an instrument is designed
Every instrument begins from a question a clinic can answer about itself using evidence it already holds. If answering a criterion requires data the clinic cannot obtain, the criterion is wrong and is rewritten. Criteria are written so that 0 and 3 are unambiguous, because a descriptor that requires interpretation produces scores that cannot be compared with the same clinic's own previous run.
Criterion referencing
No instrument scores a clinic against other clinics. Bands describe an operation against written descriptors. This is a methodological position rather than a courtesy: comparison requires a distribution, and no reliable public distribution exists for the practices assessed here.
Statistics and benchmarks
We do not invent a statistic, an industry average, a benchmark, a conversion rate, a response time or a study. Where a reader would expect one and none reliably exists, we say so in those words and explain what to do instead, which is usually to measure your own figure and compare it only with itself over time. Figures circulated in this sector are frequently supplied by organisations selling something, derived from unstated samples, and defined differently from one another.
Sources
We cite institutional sources only, and only ones we are confident exist. We do not cite a study we cannot identify, and we do not cite a secondary summary as though it were the source. Where guidance is likely to change, we name the body rather than deep-linking to a document that may move.
Commercial separation
No organisation can pay to appear in an instrument, to influence a criterion or a band, to have an instrument written, or to be named as recommended. We publish no ranking or recommendation of any supplier, agency or consultant. The sponsor slot is labelled, priced publicly and structurally incapable of touching content, and the rules governing it are on the funding page.
Estate links
One article on this site carries a single editorial link to an external organisation which is a client of our publisher. It is labelled with a disclosure on the page. It was chosen by the editor, it was not sold, and it was not paid for. It is the only link of its kind on the entire site, and every other page carries a block stating that it contains no commercial links.
Language
British English. Plain terms in preference to jargon, and technical terms explained at first use. We describe the body neutrally and do not construct a concern in order to address it, which is a standard we apply to ourselves as well as scoring it in the tone of voice instrument.
AI use
We use AI tools in research and drafting. Every instrument is reviewed by a human editor who is accountable for its accuracy, its criteria and its limits. We do not publish unreviewed generated material, and we do not publish a criterion nobody has tried to score.
Corrections
Errors are corrected promptly and noted on the instrument with the date. Where a criterion changes materially, the change is described, because a clinic that scored against the previous version needs to know whether its result still holds.