A/B testing

Test two versions of a template in a campaign and pick the winner by open rate.

An A/B test splits a campaign's sends between two versions of the same template and measures which performs better. MailingCore attributes opens and clicks per variant and picks the winner when you close it.

Model

Entity	What it is
AbTest	The test: template, status, `trafficSplit` and the winning variant
AbVariant	Each version (`versionId`) with its `sentCount`, `openCount`, `clickCount` counters

Statuses: DRAFT → RUNNING → COMPLETED (or CANCELLED).

Endpoints

Method	Path	Description
`GET`	`/ab-tests`	List tests with the template name and each variant's subject
`POST`	`/ab-tests`	Create a test (starts in `DRAFT` with two variants)
`POST`	`/ab-tests/:id/start`	`DRAFT` → `RUNNING`
`POST`	`/ab-tests/:id/complete`	`RUNNING` → `COMPLETED`, sets the winner

Guards: JwtAuthGuard + TenantGuard (dashboard authentication).

Create a test

POST /ab-tests
Authorization: Bearer <accessToken>
X-Tenant-Id: {tenantId}
Content-Type: application/json

{
  "name": "Onboarding subject",
  "templateId": "tmpl_123",
  "variantAId": "ver_a",
  "variantBId": "ver_b"
}

Field	Required	Notes
`name`	Yes	Test name
`templateId`	Yes	Template under test (must belong to your tenant)
`variantAId` / `variantBId`	Yes	The two versions to compare
`trafficSplit`	No	Stored; today both variants are created at weight 50/50

How it runs

Start the test (/start) to move it to RUNNING.
Send a campaign with that template. For each recipient, MailingCore picks a variant with a weighted split (cryptographic RNG) and increments its sentCount.
Tracking attributes the first open and the first click of each email to the matching variant (openCount / clickCount).
Complete the test (/complete): the winner is the variant with the highest open rate (openCount / sentCount); ties go to the first variant.