What the Dashboard Does, and What Is API Only
The table lists each capability and where you do it today:
- Dashboard: a screen in the dashboard does it.
- API only: it works through the API, with no screen yet. The guide in the last column shows the requests.
| Capability | Where you do it | Guide |
|---|---|---|
| Create an experiment, with guided setup or the single-page form | Dashboard | Running experiments |
| Start, pause, resume and complete an experiment | Dashboard | Running experiments |
| Change an experiment's targeting while it is a draft or paused | Dashboard | Running experiments |
| Create a feature flag | Dashboard | Creating feature flags |
| Turn a feature flag on or off | Dashboard | Creating feature flags |
| Set a flag's targeting rules and rollout percentage | Dashboard | Creating feature flags |
| See a flag's rollout schedule and its stages | Dashboard | Gradual rollouts |
| Roll a feature flag back from the safety page | Dashboard | Safety monitoring |
| Set a feature flag's per-flag safety config (error-rate and latency thresholds, rollback percentage) | API only | Safety monitoring |
| Manage users and their roles | Dashboard | Authentication user guide |
| Change your own password (local sign-in) | Dashboard | Authentication user guide |
| Read the audit log, and download it as CSV or JSON | Dashboard | Audit logging |
| Give each variant a configuration (a JSON payload) when creating an experiment | Dashboard | Running experiments |
| Turn on Bayesian analysis when creating an experiment | Dashboard | Bayesian analysis |
| See on the experiment page whether Bayesian analysis is on | Dashboard | Bayesian analysis |
| Segments: list, create, and upload a list of user ids | Dashboard | Segments |
| The sample-ratio check on the results page | Dashboard | Running experiments |
| Bayesian results: chance to be best, expected loss and credible intervals | Dashboard | Bayesian analysis |
| See a bandit experiment's current traffic weights | Dashboard | Multi-armed bandits |
| Clone an experiment | Dashboard | API endpoints |
| Delete a draft experiment | Dashboard | API endpoints |
| Edit a draft experiment's name, description and hypothesis | Dashboard | API endpoints |
| Change a draft experiment's variants or metrics | API only | API endpoints |
| Schedule an experiment's start and end dates | API only | Running experiments |
| Create or edit a rollout schedule and its stages | API only | Gradual rollouts |
| Activate, pause or advance a rollout schedule | API only | Gradual rollouts |
| Delete a feature flag | API only | API endpoints |
| Archive and unarchive a feature flag | API only | Running experiments |
| Create a bandit experiment | API only | Multi-armed bandits |
| Set a bandit's weights by hand | API only | Multi-armed bandits |
| Global holdouts: create, activate, and read their results (beta) | API only | Mutual exclusion groups and global holdout |
| Mutual exclusion groups | API only | Mutual exclusion groups and global holdout |
| Split-URL experiments (full profile) | API only | Split URL testing |
| CUPED-adjusted results (beta) | API only | CUPED |
| The interaction scan and the pair test (beta) | API only | Interaction detection |
| Download results as CSV | API only | Data export |
| The effect size in the metric comparison table | API only | Running experiments |
| Each variant's confidence interval on the results page | API only | Running experiments |
| Bayesian priors, loss threshold and credible level | API only | Bayesian analysis |