C Annex C: Measurement data types used in product quality measures

What this heading covers

The five types are the grammar of a quality measure. A count says how many; a time says how long; a ratio compares two quantities; a percentage expresses a proportion of a whole; a statistical type summarises a distribution. The choice is not cosmetic. A count cannot be averaged across products of different sizes, a percentage hides its denominator, a mean hides the tail, and a ratio of two ratios is a different claim from either. The annex is informative, but a team that ignores the type will produce reports that look precise and cannot be compared.

Record position: annex 3 of 5. The heading is C in the public contents; its text is informative and not public.

What the heading asks of a team

What to put in place before publication

1. Record the data type of every measure in the register.
2. For a ratio or a percentage, write the numerator, the denominator and the population in words.
3. For a statistical measure, name the statistic and the reason for choosing it.
4. Keep the raw values behind any summary, so the distribution can be re-read.
5. Refuse to average counts across products of different sizes without a normalising element.

Where teams usually fall short

• A count averaged across products, producing a number with no meaning.
• A percentage of a population of five, reported to two decimal places.

What happens next

Annex D applies the measures at different lifecycle stages. Read it if you need to decide when a measure can be taken.