Facts

A fact is the answer to one reported question for one entity in one year, as established by the Net Purpose data collection process. The fact itself carries the final standardised value; the evidence behind it lives in supporting_facts.

Facts are split across two views with identical shapes: revenue_facts holds the revenue-breakdown questions (product and geographic revenue segments) that feed the Revenue & SDI dataset, and outcomes_facts holds everything else.

Each fact has at least one supporting fact. Where a given fact has multiple facts the total value is given by the aggregation of the fact.

Aggregation	Description
`NONE`	There will be a supporting fact listed as `is_total`. This fact represents the aggregate level, other supporting facts are provided for context only.
`SUM`	The fact value is the sum of all the underlying segments.
`AVG`	The fact value is the average (mean) of the underlying segments.

All applicable questions are attempted to be collected for all entities in the Net Purpose universe. Where no data can be found a NOT FOUND fact record is created. The absence of a fact for a given enitty_id, question_id, year indicates that the question is not applicable to that entity or the data has not yet been looked for.

All views on this page except segments are versioned — filter is_current = TRUE (see Conventions).

`revenue_facts` and `outcomes_facts`

One row per fact, per version. Both views share the same columns:

Column	Type	Description
`fact_id`	NUMBER	Primary key
`entity_id`	NUMBER	Who. Joins to `entities`
`question_id`	NUMBER	What. Joins to `questions`
`canonical_question_id`	VARCHAR	The question’s readable slug, for convenience
`year`	NUMBER	Fiscal year of the answer determined by our assignment logic
`start` / `end`	DATE	Exact period the value covers
`value`	FLOAT	The final standardised value
`unit`	VARCHAR	Unit of the value. As a string, join to units on `units.name`
`aggregation`	VARCHAR	How supporting values were combined into the final value (e.g. `SUM`)

`supporting_facts`

The individual pieces of evidence behind a fact — typically the numbers as they appear in the source document, one row each.

Column	Type	Description
`supporting_fact_id`	NUMBER	Primary key.
`fact_id`	NUMBER	The fact this evidence supports
`recorded_value`	FLOAT	The value exactly as printed in the source
`recorded_unit`	VARCHAR	Unit of the recorded value, as printed, as a string.
`standardized_value`	FLOAT	The value after unit/currency standardisation
`standardized_unit`	VARCHAR	Unit of the standardised value, as a string
`is_total`	BOOLEAN	Whether this row is a disclosed total (rather than a component)
`is_derived`	BOOLEAN	Whether the value was derived rather than read directly. See `comment` for any derivation notes.
`is_low_confidence`	BOOLEAN	Flagged where the disclosure was ambiguous
`segment_id`	NUMBER	A descriptor of the segregation. THe type of disaggregatino depends on the metric. e.g. product_revenue is by product segments, Scope 1 CO2e emissions are by geography. Joins to `segments`
`source_id`	NUMBER	Document the value was found in. Joins to `sources`
`page_num`	NUMBER	Page of the document
`comment`	VARCHAR	Analyst note giving context on the evidence.

`segments`

Segments are disaggregations of the top-level fact. The dimension of the disaggregation depends on the question, but typically is geographic. The segment_category gives high level grouping of the different types of segment.

A segment name has three parts, from general to specific: segment_category : segment_class : segment_instance.

Column	Type	Description
`segment_id`	NUMBER	Primary key
`segment_category`	VARCHAR	Top-level grouping of the segment
`segment_class`	VARCHAR	Middle grouping
`segment_instance`	VARCHAR	The specific segment as disclosed
`type`	VARCHAR	Segment kind; most rows are plain `segment`

Example: trace a value to its evidence

SELECT f.entity_id,
       f.canonical_question_id,
       f.year,
       f.value          AS final_value,
       f.unit,
       sf.recorded_value,
       sf.recorded_unit,
       s.name           AS source_document,
       s.url,
       sf.page_num
FROM NET_PURPOSE.FOUNDATIONS.OUTCOMES_FACTS f
JOIN NET_PURPOSE.FOUNDATIONS.SUPPORTING_FACTS sf
  ON sf.fact_id = f.fact_id AND sf.is_current = TRUE
LEFT JOIN NET_PURPOSE.REFERENCE.SOURCES s
  ON s.source_id = sf.source_id
WHERE f.is_current = TRUE
  AND f.entity_id = 12345
  AND f.canonical_question_id = 'scope_1_co2e_emissions'
  AND f.year = 2024;

Facts

revenue_facts and outcomes_facts

supporting_facts

segments

Example: trace a value to its evidence

`revenue_facts` and `outcomes_facts`

`supporting_facts`

`segments`