DatasetField
One field a dataset exposes.
type DatasetField {
key: String!
label: String!
role: DatasetFieldRole!
type: DatasetFieldType!
semantic: String
aggregation: String
formula: String
group: String
description: String
definition: String
synonyms: [String!]
conformed: Boolean
computed: Boolean!
drillable: Boolean!
partOf: [String!]!
measure: DatasetMeasure
undated: Boolean!
additive: Boolean!
values: [DatasetFieldValue!]
}
Fields
DatasetField.key ● String! non-null scalar
Stable identifier used when requesting this field. Does not change once published.
DatasetField.label ● String! non-null scalar
Human-readable name.
DatasetField.role ● DatasetFieldRole! non-null enum
DatasetField.type ● DatasetFieldType! non-null enum
DatasetField.semantic ● String scalar
Hint describing what the value means, for consumers that adapt their formatting to it — for example YEAR_MONTH_DAY, YEAR_MONTH, PERCENT.
DatasetField.aggregation ● String scalar
Default aggregation for a metric: SUM, COUNT, AVG, MIN, MAX or NONE.
DatasetField.formula ● String scalar
For derived metrics such as rates, the expression to compute them from other fields — for example "SUM(opens) / SUM(sends)". Derived fields are never returned as data: a stored ratio cannot be re-aggregated, so a consumer that rolled rows up would display a wrong number. Compute it from this instead.
DatasetField.group ● String scalar
Fields sharing a group are presented together.
DatasetField.description ● String scalar
DatasetField.definition ● String scalar
The catalog's one-sentence definition of the field: what the metric counts or what the dimension groups by, as the code computes it. Null when the catalog has no entry for the field.
DatasetField.synonyms ● [String!] list scalar
Other names a reader might use for the field, from the catalog.
DatasetField.conformed ● Boolean scalar
True when the field carries a conformed dimension: one that means the same thing in every dataset that carries it, so rows from two datasets can be joined on it.
DatasetField.computed ● Boolean! non-null scalar
True when the metric is produced in PHP rather than by the fact query — a multi-touch attribution figure, say. Such a metric cannot be re-aggregated by the consumer; ask the dataset for the grain you need.
DatasetField.drillable ● Boolean! non-null scalar
True when the metric can be opened into the contacts or records behind it
with dashletDrilldown: a count of events or of contacts on a dataset that
declares a drill grain, or a sum of money on one that lists its records
(an opportunity's amount, attributed amount). False for dimensions, rates
and other computed metrics.
DatasetField.partOf ● [String!]! non-null scalar
Every metric this one is a later step of, as a funnel, nearest first: Opens is [delivered, sends], Conversions is [impressions]. A chart of measures that form such a chain draws each step inside the one before. A step counted as events (repeat opens) can be larger than its whole on a given row. Empty for a metric that is no step.
DatasetField.measure ● DatasetMeasure enum
FLOW or STOCK for a metric (see DatasetMeasure); null for a dimension. A STOCK metric cannot be drawn as a cumulative (running-total) dashlet. Nor can a FLOW whose aggregation is not SUM, or a rate: read aggregation for additivity.
DatasetField.undated ● Boolean! non-null scalar
True for a metric that means something only for the whole date range, never per date bucket (the contacts dataset's Marketable Share of Database is the share at the end of the range). It cannot be grouped by a date dimension, nor drawn as a NUMBER's trend line. False for every other field.
DatasetField.additive ● Boolean! non-null scalar
True for a count or sum whose groups add up to the whole (a SUM metric that is not derived, computed or a stock), so the groups past a chart's top N can be added into one "Other". False for a unique count, a rate, a computed metric, a stock and every dimension. Judged without a dashlet's filters: a metric that reads as a level only under some filters is still true here.
DatasetField.values ● [DatasetFieldValue!] list object
Every value a coded dimension holds, each with the name a reader knows it by, for a dimension whose rows cannot be trusted to show them all (email engagement's Action Source Type counts most senders only when a filter names them). A filter picks from these. Null for every other field.
Member Of
DashletDrilldownResponse object ● Dataset object ● DatasetRowsResponse object ● PublicDashboardDataset object


