Set Causal Settings

Set a campaign's causal-analysis configuration: the optimisation objective the uplift model targets and the ml-causal meta-learner that estimates incremental CATE on experiment conclusion. Both fields optional — a nil field leaves the stored value untouched. Pure config write; touches no lifecycle status.

POST/api/v1/marketing/dispatch

Request

curl -X POST "<BASE_URL>/api/v1/marketing/dispatch" \
  -H "Authorization: Bearer <ACCESS_TOKEN>" \
  -H "X-CQRS-Action: SET_CAUSAL_SETTINGS" \
  -H "Content-Type: application/json" \
  -d '{
    "campaign_id": "7c9e6679-7425-40de-944b-e07fc1f90ae7",
    "tenant_id": "7c9e6679-7425-40de-944b-e07fc1f90ae7"
  }'

Authorization

Minimum role: ManagerThe caller must hold this role or a higher one.
Action: SET_CAUSAL_SETTINGS

Headers

AuthorizationstringRequired
Access token of the signed-in user, as a Bearer token. It identifies the user, their role and the workspace.
X-CQRS-ActionstringRequired
Names the command to run. Must be SET_CAUSAL_SETTINGS for this endpoint.
Content-TypestringRequired
Always application/json.
X-Idempotency-KeystringOptional
A key you choose that is unique to this attempt. A command repeated with the same key is not applied a second time.
X-Correlation-IdstringOptional
Your own identifier for the request. Returned in the x-correlation-id response header; generated when you leave it out.

Request body

campaign_iduuidRequired
No description yet.
tenant_iduuidRequired
No description yet.
causal_objectiveenum or nullOptional
Optimisation target for the uplift model. None → leave unchanged.
Allowed values
- TRADITIONAL_CONVERSION
- MAXIMIZE_INCREMENTAL_UPLIFT
- MINIMIZE_CHURN_RISK
- OPTIMIZE_LIFETIME_VALUE
causal_uplift_estimatorenum or nullOptional
ml-causal meta-learner. None → leave unchanged (engine default = Auto).
Allowed values
- AUTO
- T_LEARNER
- X_LEARNER
- CAUSAL_FOREST
- DR_LEARNER
AUTO
Engine chooses the estimator from the data shape (recommended default).
T_LEARNER
Two-model meta-learner (separate treated/control models).
X_LEARNER
Cross-fitted meta-learner — robust with imbalanced treatment.
CAUSAL_FOREST
Generalized random forest for heterogeneous effects.
DR_LEARNER
Doubly-robust learner — combines outcome + propensity models.

Response

Status 202. The command was applied. The body reports the outcome and, where the command changes a record, that record's new state.

successbooleanRequired
true when the command was applied.
error_messagestringRequired
Empty when the command was applied.
projected_stateobjectOptional
The record as it stands after the command.
event_payloadobjectOptional
The event the command recorded.
insightsarray of objectOptional
Advisory notes about the result. Each has a code, a message and an impact number.

Errors

Every error carries an HTTP status. Where it also carries a code, branch on the code, not on the message.

400X-CQRS-Action is missing or names a command this route does not accept. Send SET_CAUSAL_SETTINGS.
400DOMAIN_ROUTINGThe body is not a valid command. Check the required fields and their types.
401The access token is missing, expired or not valid.
401L1_AUTHThe caller's role is below manager or the caller has no access to this record.
403L1_POLICYA business rule refused the command. The message names the rule.
409L1_IDEMPOTENCYA command with this idempotency key was already applied.
409CONFLICTThe command would duplicate a record that must be unique.
500L3_COMMITThe command was valid but could not be saved.