Failure forecast

What this feature is for

Failure forecast estimates time-to-breach from observed signals such as provider error budget, usage, or reliability drift. The important distinction is that PreFlight stores the observation alongside the app, release, provider, or customer path it belongs to. That context is what turns a signal into an operational decision.

Before you open it

A signal with enough history, a threshold that means something to the business, and a human owner for the decision.

Use a real app in the workspace and start with the smallest useful scope. A narrow, representative check gives you a stronger baseline than connecting every environment and every provider at once.

Run the workflow

Choose the signal, review the horizon and confidence, inspect the inputs, and create a preventive action before the threshold becomes an incident.

1. Open the feature from the dashboard outcome hub.
2. Choose the app and environment that match the decision you are making.
3. Run or save the configuration, then wait for the evidence state to settle.
4. Assign the next action to an owner before leaving the page.

Read the evidence

Evidence includes the measured trend, threshold, forecast window, last updated time, and model inputs. A green state means the configured assertion passed at the recorded time. It does not claim that every unconfigured path is healthy. A warning is a useful lead; a failure should have a concrete owner and a verification run.

When the result is not healthy

A short horizon with low confidence is a prompt to investigate, not permission to panic. Improve sampling and verify the source data.

Do not erase the failed result after fixing the underlying system. Keep the history, rerun the check, and link the new passing evidence to the deployment or incident that caused the change. This gives the team a useful before-and-after record.

Use it with the rest of PreFlight

Use Failure forecast with Provider SLA, Usage, Health, and Automations. Outcome hubs are deliberately connected: Audit finds risk, Deploy decides whether to promote, Health watches the running system, Revenue verifies the money path, Growth manages discoverability, and Trust publishes safe proof. Use the related links below to move between those decisions without losing context.