qbrixqbrix

Experiments in the Console

The Experiments page in the console is the primary surface for the day-to-day work of running bandit experiments. From here you can create new experiments, monitor active ones, drill into per-arm performance, and tweak feature gates without touching the API.

Experiments list view

The List View

The experiments table shows every experiment in your workspace with a few columns:

  • Experiment — the name you gave it.
  • Pool — which pool of arms it's testing.
  • Policy — the bandit algorithm. auto for meta-bandits, otherwise the concrete policy class (e.g. BetaTSPolicy).
  • Arms — the number of variants in the pool.
  • Gate — On if a feature gate is attached and enabled.
  • Status — Active or Inactive. Inactive experiments stop serving the bandit until you re-enable them.
  • Created — creation timestamp.

Use the filter tabs (All / Active / Inactive) and the search box to narrow the list.

Creating an Experiment

Click Create Experiment in the top right. The dialog walks you through four steps:

  1. Basics — Name and pool selection.
  2. Reward Type — Binary, Bounded, or Continuous. This determines which policies are eligible.
  3. Policy — auto is the default recommendation. The advanced picker lets you pick a specific algorithm.
  4. Feature Gate (optional) — Rollout %, default arm, targeting rules, and schedule.

Create experiment — policy step with auto highlighted

Tip

For most use cases, sticking with the default auto policy is the right call — qbrix will figure out which algorithm works best on your data. See the Auto Policy page for details.

The Detail View

Clicking an experiment opens its detail page, which shows:

  • Live arm-selection share — what fraction of recent traffic each arm is receiving.
  • Per-arm reward stats — pulls, average reward, conversion rate.
  • Policy parameters — the current state (Beta priors, weight vectors, etc.).
  • Feature gate config — rollout, rules, schedule, and a toggle to enable/disable.
  • Audit trail — who created it, when it was last updated.

Editing & Disabling

  • Pause an experiment by toggling Status to Inactive. Selections fall through to the gate's default arm (or error out if no gate is configured).
  • Update the gate at any time — rollout %, rules, and schedule are editable without touching the experiment itself.
  • Delete an experiment from the detail view's overflow menu. This is permanent; selection requests for the deleted ID will 404.

Next Steps