Anatomy of a rule

An AdOps rule is one document with five moving parts: a name and the ad accounts it covers, pre-filters that narrow which campaigns are in scope, a level, a list of tasks that each pair a condition set with one action, and a schedule. The builder writes all five from a single page under the header Create Rule.

Card 1 — Basic Information#

Three fields.

Rule Name carries the hint “Choose a descriptive name for your automation rule” and the placeholder “e.g., Pause underperforming campaigns”. Name rules after what they do; the name is what the Rule Logs sidebar, the Activity Log and the bulk bar all show you.

Ad Accounts is a set of removable chips plus a + button, hinted “Select up to 5 ad accounts you want to manage”.

Notifications carries the hint “Get alerts about triggered actions via email” and holds an Email Address, a Recipient Name (placeholder “e.g., Marketing Team”) and a Channel Type of Email.

Card 2 — Filter Configuration#

This card decides which campaigns the rule even looks at, before any metric is read.

A level button toggles between Campaigns and Ad sets. Beside it, a + button adds a filter chip on one of three fields:

Field	Operators	Values
Campaign ID	equal, not equal	a campaign id
Campaign Name	contains, does not contain	free text
Campaign Status	is, is not	ACTIVE or PAUSED

Each chip renders as <Field> <operator> <value> and opens an inline editor when clicked. Underneath, a live readout shows Estimated match: N Campaigns, recalculated 500 milliseconds after any change, with a per-ad-account breakdown in its tooltip. Before you attach accounts it reads “Select ad accounts to see affected items”.

Use it. It is the only place in the product that tells you the blast radius before you save.

Card 3 — Tasks & Conditions#

The toolbar holds Templates, Add Task and Generate with AI.

Each task card shows the action icon, the action name with the word “campaign” swapped for your chosen level, a one-line summary such as “Increase budget for all campaigns”, copy and trash buttons, an AND/OR pill, and a table of conditions. Every condition row is four pickers and an input: metric, period, operator, value — plus a Compare With cog and a remove button.

The stored shape is plain JSON:

{
  "level": "campaign",
  "status": "ACTIVE",
  "tasks": [
    {
      "action": "increase_budget_campaign",
      "action_frequency": "2_hours",
      "action_value": { "value_1": 100000, "value_2": 1000000, "value_3": "value" },
      "condition": {
        "operator": "AND",
        "conditions": [
          { "metric_key": "purchase_roas.omni_purchase", "period": "today", "operator": "GREATER_THAN", "value": "3", "compare": "value" },
          { "metric_key": "spend", "period": "today", "operator": "GREATER_THAN", "value": "0", "compare": "value" }
        ]
      }
    }
  ]
}

For every campaign in scope, the engine walks the tasks in order and evaluates each task’s own conditions before running that task’s own action. One rule can therefore encode a whole playbook — scale on good ROAS, pause on bad ROAS, tag either way — and each task is logged separately with its own task_id.

Card 4 — Schedule Configuration#

Either Run every <interval> or Run on spesific days and time, optionally bounded by Run within a date range. On a saved rule an info alert reports the Next Scheduled Execution. See Scheduling a rule for the full detail.

Draft or live#

The header has two buttons: Save as Draft stores the rule with status DRAFT, and Set Live stores it as ACTIVE. Only ACTIVE rules are selected by the execution engine, so a draft is a safe place to leave work in progress. Editing an existing live rule turns the primary button into Update Rule.