Create Experiment
Creates an experiment for a change you or the customer made. The change is measured on one catalog metric, before and after against a control, exactly like an applied insight: the next hourly worker pass (minute 43 UTC) freezes the 56-day baseline before changed_at, picks the control cohort from baseline data and starts measuring. No data for the scope in the baseline cancels it with no_scope_data. Experiments created without an insight carry no money.
Give metric_key with a matching scope ({page_url} for landing_* metrics, {platform, account_id, campaign_ids} for ad_* metrics) and a name. Or give insight_id: missing fields are taken from the insight, which is then marked applied at changed_at. changed_at defaults to now and must lie within the last 30 days.
The request answers experiment_conflict (409) when a live experiment already measures the same metric and scope, when the workspace has 50 live experiments created through the API, or when the insight's current application is already measured.
Requires the insights:write scope; read:* does not grant it.
API key as Bearer token
In: header
Request Body
application/json
Responses
The created experiment, status starting
/experimentscurl -X POST "https://d1iwal3igywepk.cloudfront.net/experiments" \ -H "Content-Type: application/json" \ -d '{}'{}{ "id": "497f6eca-6276-4993-bfeb-53cbbbba6f08", "name": "string", "description": "string", "source": "insight", "insight_id": "string", "ads_platform": "string", "ads_proposal_id": "string", "kind": "landing_page", "method": "before_after", "metric_key": "string", "scope": {}, "fix_type": "string", "status": "starting", "cancel_reason": "string", "verdict": "string", "evidence_grade": "A", "changed_at": "string", "changed_at_source": "string", "est_end_at": "string", "baseline_days": 0, "floor_days": 0, "cap_days": 0, "predicted_improvement_low_pct": 0, "predicted_improvement_high_pct": 0, "predicted_spend_change_pct": 0, "predicted_money_low_cents": 0, "predicted_money_high_cents": 0, "predicted_money_currency": "string", "stop_requested_at": "string", "refresh_requested_at": "string", "closed_at": "string", "created_at": "string", "updated_at": "string", "changed_via": "string", "snapshot": { "computed_at": "string", "verdict": "string", "effect": { "estimate_pct": 0, "ci_low_pct": 0, "ci_high_pct": 0, "undefined_reason": "string", "spend_change_pct": 0 }, "money": { "estimate_cents": 0, "low_cents": 0, "high_cents": 0, "currency": "string" }, "sample": { "pre_days": 0, "post_days": 0, "required_post_days": 0, "phi": 0 }, "placebo": { "estimate_pct": 0, "ci_low_pct": 0, "ci_high_pct": 0, "failed": true }, "series": [ { "date": "string", "target_num": 0, "target_den": 0, "control_num": 0, "control_den": 0, "settled": true } ] }}