Skip to content

Experiments Overview

Experiments are the core of Ttoolab. They let you test changes, measure impact, and decide what to ship.

sequenceDiagram
participant Browser
participant Pixel as Ttoolab Pixel
participant Edge as SDK Edge
participant Dashboard as Ttoolab Dashboard
Dashboard->>Edge: Publish experiment config
Browser->>Pixel: Page load
Pixel->>Edge: GET /sdk/bootstrap
Edge-->>Pixel: Assignments + variant changes
Pixel->>Browser: Apply DOM changes
Pixel->>Edge: POST /sdk/event ($exposure)
  1. You create and configure an experiment in the dashboard.
  2. You publish the experiment. Ttoolab stores the configuration on its edge network.
  3. When a visitor loads a matching page, the pixel calls /sdk/bootstrap.
  4. The edge service checks URL rules, traffic allocation, and targeting, then assigns a variant.
  5. The pixel applies variant changes (for A/B tests) and sends an exposure event.
  6. Conversions are recorded when goal conditions are met or when you send matching custom events.

| Type | Description | Doc | | :--- | :---------- | :— | | A/B test | Modify DOM elements on the same page | A/B Tests | | Split URL test | Route visitors to different page URLs | Split URL Tests | | Redirect test | Redirect visitors to a destination URL | Redirect Tests | | Feature flag | Toggle code paths by variant assignment | Feature Flags |

Every experiment defines which pages it runs on using URL rules:

| Match type | Behavior | | :--------- | :------- | | contains | Current URL includes the pattern | | exact | Current URL equals the pattern | | starts_with | Current URL starts with the pattern | | regex | Current URL matches a regular expression |

Traffic allocation controls what percentage of visitors enter the experiment (0–100%). Visitors outside the allocation are not assigned.

When sticky assignment is enabled, returning visitors keep the same variant. The edge stores the assignment keyed by anonymous ID and experiment key.

  • Publish — pushes the experiment to the edge and sets status to running.
  • Unpublish / pause — removes the experiment from the active manifest. The pixel stops assigning new visitors.

Webhook events experiment.published and experiment.started fire when you publish.

Ttoolab tracks:

  • Exposures — visitors assigned to a variant (automatic)
  • Conversions — visitors who meet a goal condition

View results in the Ttoolab dashboard. Optionally push data to Google Analytics or receive webhook notifications.

  • Test one major change per experiment when possible.
  • Define a primary goal before launching.
  • Run experiments long enough to reach meaningful sample sizes.
  • Avoid changing variant content after launch without creating a new experiment version.