11. Outputs and Interpretation

11.1 Primary Outputs

Each run publishes one row for every combination of data version, provider, and region, carrying the following outputs:

• Relative cost percentage, expressed as a percentage difference from the median region of the same provider. The region's catalogue price position within that provider's estate. Suppressed where basket coverage is insufficient.
• Cost grade, expressed as a value from 1 to 5. The band the relative cost percentage falls in, where 1 is the least expensive band. Suppressed with the relative cost percentage.
• Basket coverage percentage, expressed as a percentage of the comparison basket. The share of basket product categories for which the region publishes a price.
• Service coverage percentage, expressed as a percentage of the provider's most complete region. The region's catalogue completeness within that provider's estate.
• Coverage grade, expressed as a value from 1 to 5. The band the service coverage percentage falls in, where 1 is the most complete band.

11.2 Intended Use

The scores support region selection and region comparison within a single provider, for example in deciding between two candidate regions of comparable latency, or in establishing whether a region under consideration prices the services a workload needs. They also support a first read on whether a region's catalogue prices are above or below the provider's own middle, ahead of any detailed pricing exercise.

11.3 Interpretation Guidance

The percentages are the stable output and the grades are a banded summary of them. Where the two appear to disagree between versions, the percentages carry the position.

A grade cannot be converted back into a range of percentages, because the band boundaries are fitted for each provider in each version and are not published. Two regions sharing a grade may be some distance apart within the band.

The composition of the comparison basket determines what the cost score is sensitive to. Every basket product category carries equal weight, and compute product categories are the most numerous, because each machine type forms its own product category. A region whose compute prices are above its provider's middle will therefore tend towards a higher relative cost percentage even where its storage prices are below it.

A cost score published alongside a low basket coverage percentage rests on a narrower slice of the catalogue than one published alongside a high figure, and a reader comparing the two needs to allow for that.

11.4 Non-Intended Use

The scores are not a forecast or an approximation of a bill. No customer usage or spend is used in their calculation, and a customer whose consumption is concentrated in one service will not find their own cost position in a score computed over the whole basket.

Grades and percentages are not comparable across providers. A cost grade of 4 on one provider and a cost grade of 4 on another describe positions within two separate estates and carry no relationship to each other.