Evaluation Spec Filters

In Schedule Based rules, it is possible to use more advanced types of filter fields.

Prefixed Insights Fields

You can set specific types of prefixes for Insights filters. This is similar to prefixes set for Metadata filters to perform multi-level filtering.

You can set an object level prefix on a filter to leverage Insights multi-level filtering. For example, an ads rule can be filtered by ad set or ad campaign performance. You can also specify attribution window and time preset prefixes on a filter. This is used to override the rule's attribution window and time preset for this particular filter.

Usage

Prefixes are optional. A field can have:

• an object level prefix
• an attribution window prefix
• a time preset prefix

You can have all the options above, none, or any combination, as long as you keep them in that order. The field should have following format:

{ object_level_prefix? } {attribution_window_prefix?} { time_preset_prefix? } { field_name }

Below, see examples of correct and incorrect prefixed Insights fields. We also provide examples of correct and incorrect prefixed metadata fields, to show which metadata filters we support.

Examples for Insights field spent

Good examples

Bad examples

Examples for Metadata field daily_budget

Good examples

Bad examples

Object Level Prefixes

Attribution Window Prefixes

Time Preset Prefixes

This list is the same as the list of valid time preset values, but made into lowercase and attached with a delimiter.

Aggregation

You can aggregate some Insights fields over multiple ad objects. This allows you to create filters on metrics over a specific subset of ad objects. For example, the total reach over multiple ads or the total number of clicks over different ad sets.

While some metrics such as clicks are calculated with simple summation, other metrics such as reach are calculated differently. Since reach is based on unique impressions, duplicate users are removed in the aggregation over multiple ad objects.

Usage

An aggregated field takes the form of aggregate({ field }). The field can contain attribution window and time preset prefixes. The set of ad objects to be aggregated is determined by another required filter field aggregation_id.

Example Aggregate Fields

Good examples

Bad examples

Aggregation Id Filter

The aggregation_id filter specifies which ad objects to aggregate. It only supports the IN operator and a list of ids as its value. The ids may be of ads, ad sets, or ad campaigns, but all of them must be from the same object level.

Example aggregation_id filter

{
  "field": "aggregation_id",
  "operator": "IN",
  "value": [1234, 5678]
},
{
  "field": "aggregate(reach)",
  "operator": "GREATER_THAN",
  "value": 100
}

Supported Fields

• clicks
• cpc
• cpm
• cpp
• ctr
• frequency
• impressions
• mobile_app_purchase_roas
• reach
• result_rate
• spent
• unique_clicks
• unique_impressions
• website_purchase_roas
• cost_per_unique_click

Formula Fields

You can set simple arithmetic expressions as a field. For example, this can be used to find the ratio between two numeric fields.

This works on Insights fields and on a subset of numeric Metadata fields. The full supported list is below.

Usage

A formula field consists of fields or constants and syntactically correct operators separated by spaces. It supports the operators + - * and /. You can add constants, for example to weigh specific fields or to act as offsets.

The fields in this case can be fully prefixed, so you can add valid object level and time preset prefixes.

We currently allow a maximum of 6 non-constant fields in a formula. You can have as many constants as you want.

today_spent / adset.today_spent

0.8 * cpc + 0.2 * cpm

{ field_or_constant_1 } { + | - | * | / } { field_or_constant_2 }

Example Formulas

Good examples

Bad examples

Valid Numeric Metadata Fields