meta_create_test
Commit to a Next test: it becomes a Test whose intent is FROZEN — hypothesis, reference, keep / change, funnel, goal, and the rule that will judge it (comparator, the expected volume it needs before judging, how long it waits for late conversions, how long it runs) — with the evidence it was decided on.
write meta_create_test
Commit to a Next test: it becomes a Test whose intent is FROZEN — hypothesis, reference, keep / change, funnel, goal, and the rule that will judge it (comparator, the expected volume it needs before judging, how long it waits for late conversions, how long it runs) — with the evidence it was decided on. Pass the Next test's nextTestId with the same window / funnel it was computed in; or, for an idea no Next test proposed, custom with its title, hypothesis, what to try and the evidence it rests on. Comparator baseline (default) holds the test to the cost per goal the decision was made on; reference races it against the reference creative's ads over the same days — only if both run. A changed mind is a new test with supersedesId, never an edit. The brief it judges (briefId, or the Next test's saved brief) is linked to it and becomes learned when it concludes. Touches nothing on Meta. A Test spends nothing: its ads are launched PAUSED with its testId, and its outcome is recorded as Observations.
Parameters
| Parameter | Type | Required | Description |
|---|---|---|---|
nextTestId | string | No | |
window | 30d | 90d | 180d | lifetime | No | |
funnel | instant_form | website | messaging | profile | traffic | unknown | No | The funnel the Next tests were read in; for an exploration that pools funnels, where to run it. |
custom | object | No | |
hypothesis | string | No | Overrides the default hypothesis written from the Next test. |
comparator | reference | baseline | No | |
minExpectedGoals | number | No | |
maturityDays | integer | No | |
windowDays | integer | No | |
supersedesId | string | No | |
briefId | string | No | The brief this Test judges; it is linked, and learned when the Test concludes. |
get_testread
One Test in full, by its id: its frozen intent (hypothesis, reference, keep / change, funnel, goal, the rule that judges it), the evidence snapshot it was decided on, what it produced (assets, ads, launches) and every Observation of its outcome, newest first.
link_test_artifactswrite
Tie what a Test produced to it, when it was not launched with its `testId`: assets (uploads), Meta ad ids, launches.